Educational companion dossier · Fact, interpretation, lived experience, clinical education, fiction, and mechanics are labeled separately. Scope & safety
PsychologicalWar.orgHistory before hype.

AI PSYOPS · FIELD REALISM · WIP.69

From plausible claim to bounded finding

A defensive correction-and-currentness layer for deciding what changed, what remains disputed, and what an incident, model, law, appeal, forecast, authentication signal, or bounded search actually establishes.

The twelve-category taxonomy describes how AI can intersect with influence. WIP.69 advances that research foundation with formal schema-v2 and schema-v1 export conformance, immutable release deltas, fail-closed review-queue recovery, public correction observability, external request and receipt accountability with UNKNOWN_NOT_INFERRED, cross-runtime test-only signature regression, synthetic-only human execution-kit conformance that remains unexecuted, and selective currentness version 12 while continuing to follow the evidence across versions: from prior wording to corrected wording, from source overlap to contradiction, from artifact to effect, from appeal to downstream repair, from sensor to construct validity, and from predictive output to the institutional action taken on it.

Open claim and correction status Review correction research Compare predictive deployments Compare search versions Return to the taxonomy
REAL-WORLD INTERPRETIVE

A CROSS-CUTTING EVIDENCE LAYER

Exact owner reports, bounded authority records, and visible unresolved evidence

WIP.64 retains the twelve exact field-realism owner-report records and six exact WIP.63 research foundations, then adds live claim review operations, public correction status, fail-closed correction transactions, privacy controls, and currentness version 7. It versions 65 authority records, 13 claim reviews, 13 correction and supersession records, 13 contradiction and origin-chain records, 56 source-quality records, 45 bounded-search records including 9 WIP.62 version-5 refreshes, 9 high-consequence evidence-gap reviews, and 7 longitudinal predictive-deployment records. The predictive-population deployment report remains visibly repository-authored because no dedicated owner report on that topic was supplied in the WIP.55 intake.

What this changes: public wording now exposes prior wording, current wording, changed fields, reason for change, source agreement or conflict, source independence, longitudinal status, prior and refreshed search IDs, known blind spots, correction triggers, and reopening triggers.

NOT_RETRIEVED: the recorded search did not locate the evidence. The label does not mean that the evidence never existed, was destroyed, or is impossible to obtain.

What this does not change: source review is not a legal opinion, clinical diagnosis, scientific certification, intelligence attribution, affected-community endorsement, accessibility certification, publication approval, production authority, or permission to conduct psychological operations.

SIX REALISM GATES

Do not let one observation silently become a stronger conclusion

Each gate prevents a common analytical collapse. A source can pass one gate and still leave every later stage unresolved.

  1. Artifact → effect

    An artifact can exist without proving coordination, actor identity, sponsorship, intent, recipient belief, behavior change, or strategic effect.

  2. Appeal → repair

    A reversal or restored item can stop an immediate error without correcting copied data, derived scores, rankings, vendor records, lost income, reputation, or repeated harm.

  3. Signal → mind

    A facial movement, vocal feature, gaze pattern, physiological signal, or interaction trace is not direct knowledge of emotion, deception, loyalty, vulnerability, or intent.

  4. Rule → universal law

    One enacted rule, judgment, settlement, guidance document, or proposal does not become a universal cognitive-liberty code.

  5. Account → prevalence

    An attributable first-person or institutional record can document harm without establishing population prevalence, one exclusive cause, or community endorsement.

  6. Indicator → attribution

    A generated face, detector score, linguistic pattern, or synchronized post is not enough by itself to establish a coordinated network, actor, sponsor, intent, or effect.

TWENTY-STAGE EFFECT ATTRIBUTION

Follow the complete chain—then stop where the evidence stops

Distribution metrics, psychological outcomes, operational outcomes, and strategic effects are separate findings. A sponsor’s objective is not a measured effect.

  1. 01Artifact or event existence
  2. 02Content status
  3. 03Coordination
  4. 04Actor identity
  5. 05Sponsorship or direction
  6. 06Intent
  7. 07Output
  8. 08Distribution
  9. 09Availability
  10. 10Reach
  11. 11Exposure
  12. 12Attention
  13. 13Recall
  14. 14Comprehension
  15. 15Credibility
  16. 16Belief or attitude
  17. 17Intention
  18. 18Behavior
  19. 19Operational outcome
  20. 20Strategic effect

Publication rule: output does not establish distribution; reach does not establish exposure; exposure does not establish credibility; credibility does not establish belief; behavior does not automatically establish operational or strategic effect.

THIRTEEN CLAIM-SPECIFIC REVIEWS

Public wording paired with the strongest source and its limits

Open a review to see the source relationship, alternatives, measurement limits, privacy boundary, currentness, and the evidence that would reopen the assessment.

  1. CLAIM-REAL-01-REPAIR Reversal is not complete repairPRIMARY_AUTHORITY_CLOSURE_PARTIAL
    A successful appeal, reversal, settlement, or restored account can stop an immediate error without repairing every copied record, derived score, ranking signal, vendor system, financial loss, reputational consequence, or psychological harm. Federal Court approval on 23 June 2026 of an additional Robodebt settlement documents compensation and remedy stages, not proof of complete downstream repair.

    Evidence supports

    Remedy must be decomposed into notice, access, authoritative review, reversal, restoration, data correction, downstream propagation, compensation, repeat prevention, and unresolved harm.

    Does not establish

    That every appealed system fails downstream repair or that every reversal is ineffective.

    Claim type
    REMEDY_OUTCOME
    Scope
    Cross-domain automated decisions and platform governance
    Source relationship
    Official inquiry plus a primary platform appellate decision illustrate distinct remedy stages.
    Currentness
    CURRENT_AS_OF_2026-07-29_WITH_STATED_LIMITS
    Specialist disposition
    NOT_COMPLETED

    Measurement limits

    Public decisions rarely disclose all internal ranking, feature-store, vendor, cache, and backup changes.

    Generalization limits

    A platform-content appeal and a public-benefits inquiry are different legal and technical systems.

    Privacy boundary

    Do not publish personal records or infer private harms beyond attributable evidence.

    Affected-person boundary

    Published testimony and institutional findings can establish experienced harms without making every account representative.

    Competing explanations

    • Some systems may fully correct a limited record even when broader harm is absent.
    • A visible restoration may be complete for a narrow content-removal dispute but not for other domains.

    Primary or authoritative records

    Correction trigger: A source shows complete downstream correction and remediation for a cited case.

    Reopening trigger: New audit, compensation, propagation, or repeat-error data becomes available.

  2. CLAIM-REAL-02-UNLEARNING Deletion and output blocking are not model unlearningPRIMARY_AUTHORITY_CLOSURE_SUBSTANTIAL
    Source deletion, suppression, output blocking, model editing, retraining, exact unlearning, and approximate unlearning are distinct interventions. Current standards and public research do not establish that a named production foundation model has exactly removed a requested influence or reached independently audited retraining-equivalent behavior across hidden and observable pathways.

    Evidence supports

    Source deletion, suppression, output blocking, retraining, and influence removal are distinct interventions.

    Does not establish

    That exact unlearning is feasible for every model or that a specific implementation satisfies law.

    Claim type
    TECHNICAL_DATA_LINEAGE
    Scope
    Machine-learning systems, retrieval systems, agents, and data pipelines
    Source relationship
    Official technical definition supported by primary research on an efficient unlearning architecture.
    Currentness
    CURRENT_AS_OF_2026-07-29_WITH_STATED_LIMITS
    Specialist disposition
    NOT_COMPLETED

    Measurement limits

    Verification requires comparing model behavior or parameters against an appropriate counterfactual and tracing downstream systems.

    Generalization limits

    Methods for sharded supervised models do not automatically transfer to foundation models, reinforcement learning, or proprietary multi-vendor systems.

    Privacy boundary

    Verification should avoid re-exposing the very personal data being removed.

    Affected-person boundary

    A person’s right to correction does not imply that every probabilistic inference can be labeled factually true or false without context.

    Competing explanations

    • For an untrained rules system, source deletion may be sufficient.
    • A model may never have ingested the corrected record.

    Primary or authoritative records

    Correction trigger: NIST or a recognized standard adopts a materially different definition.

    Reopening trigger: Validated exact-unlearning benchmarks for large deployed models become available.

  3. CLAIM-REAL-03-MENTAL-STATE Observed movement is not direct knowledge of emotionPEER_REVIEWED_CLOSURE_SUBSTANTIAL
    A system may detect a facial movement, vocal feature, gaze pattern, physiological signal, or interaction trace without reliably establishing the person’s emotion, deception, loyalty, vulnerability, intent, or dangerousness. Independent construct-specific field validity remains incomplete.

    Evidence supports

    Sensor and feature accuracy must be separated from label quality, construct validity, calibration, generalization, base rates, subgroup burden, and institutional action.

    Does not establish

    That no behavioral or physiological signal has any predictive utility for any narrow purpose.

    Claim type
    SCIENTIFIC_CONSTRUCT_VALIDITY
    Scope
    Affective computing and mental-state inference
    Source relationship
    Peer-reviewed scientific review provides the construct-validity boundary; legislation supplies a bounded governance response.
    Currentness
    CURRENT_AS_OF_2026-07-29_WITH_STATED_LIMITS
    Specialist disposition
    NOT_COMPLETED

    Measurement limits

    Ground truth for latent mental constructs is often contested, culturally dependent, and affected by context.

    Generalization limits

    Facial-expression evidence cannot be generalized automatically to every modality or population.

    Privacy boundary

    Do not treat inferred states as private facts or publish individualized mental-health conclusions.

    Affected-person boundary

    Subgroup performance, disability, neurodiversity, and cultural expression require separate evaluation.

    Competing explanations

    • Multimodal systems may improve prediction for a defined operational outcome without reading an inner state.
    • Context-specific self-report labels may support narrow research uses.

    Primary or authoritative records

    Correction trigger: Large replicated field studies establish robust construct validity for a specified system and use.

    Reopening trigger: New scientific consensus or regulatory guidance materially changes the boundary.

  4. CLAIM-REAL-04-LAW Cognitive-liberty law is jurisdiction-specific and unevenPRIMARY_LEGAL_CLOSURE_PARTIAL
    Cognitive-liberty, neural-data, biometric, automated-decision, and AI-manipulation rules remain jurisdiction-specific, staged, and uneven. A proposal, guidance document, settlement, trial judgment, or one jurisdiction’s enacted rule is not a universal code.

    Evidence supports

    Legal status must separate proposal, enactment, entry into force, application, guidance, complaint, enforcement, settlement, judgment, amendment, repeal, and supersession.

    Does not establish

    Legal advice, universal application, or uniform enforcement.

    Claim type
    LEGAL_CURRENTNESS
    Scope
    International, supranational, national, and subnational law
    Source relationship
    Primary enacted legal texts establish bounded rules in their own jurisdictions.
    Currentness
    CURRENT_AS_OF_2026-07-29_WITH_STATED_LIMITS
    Specialist disposition
    NOT_COMPLETED

    Measurement limits

    Public legal status may lag implementation, enforcement, or litigation.

    Generalization limits

    EU and Colorado rules cannot be universalized to other jurisdictions.

    Privacy boundary

    Legal summaries should not expose complainants or infer protected traits.

    Affected-person boundary

    Formal rights may remain inaccessible because of cost, language, disability, retaliation, or standing barriers.

    Competing explanations

    • Existing privacy, consumer-protection, disability, labor, or human-rights law may cover harms without using the term cognitive liberty.
    • Soft-law principles can influence practice without being binding law.

    Primary or authoritative records

    Correction trigger: A law is amended, repealed, corrected, or authoritatively interpreted.

    Reopening trigger: Implementing rules, enforcement, or appellate decisions become available.

  5. CLAIM-REAL-05-VISIBILITY Visibility actions and remedy outcomes are separatePRIMARY_DECISION_CLOSURE_PARTIAL
    Removal, restriction, search exclusion, recommendation exclusion, downranking, labeling, monetization change, account penalty, personalized refusal, synthesized reframing, and persistent-profile change are different actions. A visible restoration does not by itself show that ranking, recommendation, strikes, income, or downstream records were repaired.

    Evidence supports

    A content item can be restored without evidence that reach, recommendation, strikes, monetization, or copied signals were repaired.

    Does not establish

    That every traffic decline is suppression or every ranking change is censorship.

    Claim type
    PLATFORM_GOVERNANCE_OUTCOME
    Scope
    Ranking, recommendation, moderation, and persistent profiles
    Source relationship
    A primary appellate decision demonstrates removal, notice, appeal, reversal, and restoration as separate stages.
    Currentness
    CURRENT_AS_OF_2026-07-29_WITH_STATED_LIMITS
    Specialist disposition
    NOT_COMPLETED

    Measurement limits

    Platforms rarely reveal counterfactual distribution, recommendation coefficients, or all internal enforcement signals.

    Generalization limits

    One platform decision does not describe every platform or AI assistant.

    Privacy boundary

    Appeal transparency should not expose reporter identities, private safety data, or vulnerable users.

    Affected-person boundary

    Creator income and audience effects require attributable evidence rather than inferred reach loss.

    Competing explanations

    • Audience demand, competition, seasonality, product changes, or ordinary ranking updates can reduce reach.
    • Some interventions are lawful, transparent, and necessary for safety.

    Primary or authoritative records

    Correction trigger: Platform records show that a named action or outcome was misclassified.

    Reopening trigger: Audit data reveals downstream ranking, strike, monetization, or profile changes.

  6. CLAIM-REAL-06-AFFECTED-EVIDENCE Lived experience and institutional evidence require explicit scopeOFFICIAL_INQUIRY_CLOSURE_PARTIAL
    First-person, representative, institutional, aggregated, anonymized, and derivative evidence can document different parts of an AI-governance harm. None should be silently converted into prevalence, universal causation, representativeness, or affected-community endorsement.

    Evidence supports

    Affected-person evidence belongs in governance analysis with consent, privacy, retaliation, selection, nonresponse, and representativeness limits.

    Does not establish

    Population prevalence from a case collection or an exclusive causal chain for every reported harm.

    Claim type
    AFFECTED_PERSON_EVIDENCE
    Scope
    Workers, applicants, tenants, benefits claimants, students, creators, families, and other affected groups
    Source relationship
    Official inquiry record integrates institutional evidence and published testimony while preserving a bounded source context.
    Currentness
    CURRENT_AS_OF_2026-07-29_WITH_STATED_LIMITS
    Specialist disposition
    NOT_COMPLETED

    Measurement limits

    Nonresponse, survivorship, legal selection, and inaccessible records limit inference.

    Generalization limits

    One program, community, or jurisdiction cannot represent all systems.

    Privacy boundary

    Do not identify private people, publish sensitive records, or infer diagnosis, immigration status, disability, or trauma.

    Affected-person boundary

    No project-run interviews, consent, or community endorsement are claimed in WIP.56.

    Competing explanations

    • Administrative, economic, health, family, or institutional factors may contribute alongside automation.
    • Published testimony may overrepresent severe or legally salient cases.

    Primary or authoritative records

    Correction trigger: A cited account is corrected, withdrawn, or shown to be misrepresented.

    Reopening trigger: Authorized community research, representative surveys, or new official records become available.

  7. CLAIM-REAL-07-COMPANION-SAFETY Regulatory inquiry is not clinical causalityPRIMARY_REGULATORY_CLOSURE_LIMITED
    The FTC’s companion-chatbot work remained publicly described as an ongoing Section 6(b) study in April 2026. An inquiry documents regulatory concern and information gathering; it is not a completed finding of wrongdoing, clinical causation, or prevalence.

    Evidence supports

    Companion-system safety requires product-design, age, disclosure, data, engagement, crisis, and dependency review.

    Does not establish

    Prevalence, diagnosis, causation, negligence, or liability.

    Claim type
    CLINICAL_AND_REGULATORY_BOUNDARY
    Scope
    AI companion products and conversational systems
    Source relationship
    Primary regulatory notice defines inquiry scope without adjudicating clinical causation.
    Currentness
    CURRENT_AS_OF_2026-07-29_WITH_STATED_LIMITS
    Specialist disposition
    NOT_COMPLETED

    Measurement limits

    Case reports, media accounts, and complaints cannot establish population prevalence or comparative risk alone.

    Generalization limits

    Different products, model versions, ages, and usage patterns have different risk profiles.

    Privacy boundary

    Do not diagnose identifiable users or publish private chats without authorization.

    Affected-person boundary

    Families and users should be represented through attributable, consent-aware records with withdrawal and retaliation safeguards.

    Competing explanations

    • Users may have pre-existing vulnerabilities, sleep loss, substance exposure, grief, isolation, or other stressors.
    • Some users report short-term support without harmful dependency.

    Primary or authoritative records

    Correction trigger: FTC publishes findings or corrects the inquiry scope.

    Reopening trigger: Peer-reviewed longitudinal evidence, final enforcement, or adjudicated product records become available.

  8. CLAIM-REAL-08-INCIDENT-EFFECT An incident record does not automatically prove behavioral effectPRIMARY_AUTHORITY_CLOSURE_SUBSTANTIAL_FOR_INCIDENT
    A verified artifact, coordinated distribution, operator record, or enforcement action can establish an incident without proving recipient exposure, belief, intention, behavior change, operational outcome, or strategic effect.

    Evidence supports

    Incident analysis must separate existence, content, coordination, identity, sponsorship, intent, output, distribution, availability, reach, exposure, attention, recall, comprehension, credibility, belief, intention, behavior, operational outcome, and strategic effect.

    Does not establish

    Recipient belief, turnout change, electoral outcome, or broad AI persuasion effectiveness.

    Claim type
    INCIDENT_AND_EFFECT_ATTRIBUTION
    Scope
    Election-related synthetic audio incident
    Source relationship
    Final agency action closes artifact, operator, distribution, and enforcement stages while leaving belief and behavior unmeasured.
    Currentness
    CURRENT_AS_OF_2026-07-29_WITH_STATED_LIMITS
    Specialist disposition
    NOT_COMPLETED

    Measurement limits

    Recipient-level exposure, belief, and behavior data are not available in the cited enforcement record.

    Generalization limits

    A targeted robocall cannot represent all synthetic media or election contexts.

    Privacy boundary

    Do not publish recipient phone numbers or private voter data.

    Affected-person boundary

    Recipient experience is not inferred without attributable testimony or research.

    Competing explanations

    • Recipients may have recognized the anomaly or learned of the correction before acting.
    • Media coverage may have expanded awareness of the fake more than the original calls.

    Primary or authoritative records

    Correction trigger: FCC or a court changes the operator, scope, or final disposition.

    Reopening trigger: Survey, telecom, or election research measures exposure, belief, or behavior.

  9. CLAIM-REAL-09-SWARM-DETECTION Behavioral coordination evidence is stronger than a synthetic-content score alonePRIMARY_PLATFORM_CLOSURE_PARTIAL
    Behavioral and infrastructure evidence can establish coordination more strongly than an AI-content score alone. Coordination still does not automatically establish actor identity, sponsorship, intent, audience effect, or the synthetic origin of every asset.

    Evidence supports

    Content generation, coordination, actor identity, sponsorship, intent, and effect are separate claims.

    Does not establish

    That pseudonymous users, non-native writers, activists, role-players, or neurodivergent users are bots because of linguistic style or imagery.

    Claim type
    ATTRIBUTION_AND_FALSE_POSITIVES
    Scope
    Synthetic personas, coordinated networks, and platform threat detection
    Source relationship
    Primary platform reports document observed misuse while preserving behavior-centered enforcement and bounded reach findings.
    Currentness
    CURRENT_AS_OF_2026-07-29_WITH_STATED_LIMITS
    Specialist disposition
    NOT_COMPLETED

    Measurement limits

    Platform access and private telemetry create verification asymmetry; public researchers may lack decisive metadata.

    Generalization limits

    Detection signals vary by platform, language, community, and adversary adaptation.

    Privacy boundary

    Avoid de-anonymizing lawful pseudonymous users or publishing sensitive network identifiers.

    Affected-person boundary

    False-positive burden must be evaluated for marginalized, anonymous, and emergency-response communities.

    Competing explanations

    • Organic communities can synchronize around breaking events.
    • Marketing automation, accessibility tools, translation, or shared templates can resemble coordination.

    Primary or authoritative records

    Correction trigger: A platform retracts or materially changes a network attribution.

    Reopening trigger: Independent forensic, legal, or financial evidence establishes actor or sponsor identity.

  10. CLAIM-REAL-10-DEEPFAKE-RESPONSE Official debunking closes some stages, not all effectsPRIMARY_INCIDENT_CLOSURE_PARTIAL
    The Zelenskyy surrender deepfake and rapid rebuttal establish the artifact, hacked distribution, and authentication response. Public sources reviewed here do not establish representative recipient belief, frontline behavior change, operational outcome, or strategic effect.

    Evidence supports

    Rapid live confirmation, official channels, redundancy, platform response, and correction archives are distinct crisis controls.

    Does not establish

    Precise origin, continuing belief, behavior change, operational disruption, or strategic effect.

    Claim type
    CRISIS_AUTHENTICATION_AND_EFFECT
    Scope
    Wartime leader impersonation and institutional response
    Source relationship
    Primary targeted-state incident report documents artifacts and response; technical standard informs future provenance rather than retroactive proof.
    Currentness
    CURRENT_AS_OF_2026-07-29_WITH_STATED_LIMITS
    Specialist disposition
    NOT_COMPLETED

    Measurement limits

    Audience belief and frontline exposure are not fully measured in the official report.

    Generalization limits

    A wartime national-leader incident differs from local fraud, elections, or private-person harassment.

    Privacy boundary

    Do not expose private recipients, device logs, or operational security details.

    Affected-person boundary

    Civilian and service-member experiences require direct, consent-aware evidence.

    Competing explanations

    • Low technical quality may have limited credibility.
    • Repeated media coverage of the debunk may have increased awareness of the artifact.

    Primary or authoritative records

    Correction trigger: The issuing body or stronger forensic evidence changes chronology or attribution.

    Reopening trigger: Platform telemetry, intelligence release, or recipient research measures exposure and effect.

  11. CLAIM-REAL-11-LIARS-DIVIDEND Provenance is not truth, and the liar’s dividend is context-dependentPRIMARY_STANDARD_PLUS_PEER_REVIEWED_CLOSURE_SUBSTANTIAL
    C2PA 2.4 supports signed provenance claims, durable credentials, soft bindings, manifest recovery, and trust-list governance. Credentials can still be absent, stripped, untrusted, or semantically misleading; they do not prove the depicted event is true, and their absence does not make media false. Liar’s-dividend effects remain context-dependent.

    Evidence supports

    Authentication should combine provenance, source confirmation, live interaction, reverse search, forensics, channel redundancy, correction archives, and uncertainty disclosure.

    Does not establish

    That provenance guarantees truth or that deepfake denial always succeeds.

    Claim type
    PROVENANCE_AND_LIARS_DIVIDEND
    Scope
    Crisis authentication, signed media, and false denial
    Source relationship
    Primary standard defines what provenance authenticates; peer-reviewed experiments bound the liar’s-dividend effect.
    Currentness
    CURRENT_AS_OF_2026-07-29_WITH_STATED_LIMITS
    Specialist disposition
    NOT_COMPLETED

    Measurement limits

    Trust-list governance, key compromise, metadata loss, and user comprehension affect real-world performance.

    Generalization limits

    Survey experiments do not reproduce every high-stakes crisis or media ecosystem.

    Privacy boundary

    Provenance systems should minimize unnecessary device and creator identity exposure.

    Affected-person boundary

    Authentication should not require vulnerable witnesses to reveal identities publicly.

    Competing explanations

    • Authentic media can lack credentials because of legacy devices, screenshots, recompression, or stripped metadata.
    • Signed media can still depict staged, misleading, or selectively framed events.

    Primary or authoritative records

    Correction trigger: C2PA or the journal corrects the relevant records.

    Reopening trigger: Field evidence on credential adoption, attack resistance, or liar’s-dividend effects becomes available.

  12. CLAIM-REAL-12-DEMOCRATIC-DEFENSE Democratic defense should target behavior and foreign interference, not lawful beliefPRIMARY_GOVERNANCE_CLOSURE_PARTIAL
    Democratic cognitive defense should target concealed coordination, foreign direction, impersonation, technical tampering, artificial amplification, fraud, and platform abuse rather than lawful belief. Current VIGINUM and CNIL records document institutional activity, oversight, retention limits, and indirect access, rectification, and erasure routes—not a public register proving false-positive frequency or completed remedy outcomes.

    Evidence supports

    Mandates should be specific, legally bounded, transparent, independently overseen, correction-capable, and separated from content or belief policing.

    Does not establish

    That every institutional classification is correct or that the French model is universally transferable.

    Claim type
    GOVERNANCE_AND_CIVIL_LIBERTIES
    Scope
    Democratic information-integrity institutions
    Source relationship
    Official French threat assessment documents a foreign-interference and behavior-focused mandate.
    Currentness
    CURRENT_AS_OF_2026-07-29_WITH_STATED_LIMITS
    Specialist disposition
    NOT_COMPLETED

    Measurement limits

    Effectiveness and false-positive rates require independent oversight and published outcome data.

    Generalization limits

    Constitutional, intelligence, platform, and administrative law differs across countries.

    Privacy boundary

    Open-source analysis should not become general monitoring of lawful domestic communities.

    Affected-person boundary

    Targets of erroneous classification need notice, correction, and remedy where disclosure is legally possible.

    Competing explanations

    • Some harmful campaigns use authentic domestic actors or lawful speech, complicating mandate boundaries.
    • Foreign-origin evidence may be incomplete or politically contested.

    Primary or authoritative records

    Correction trigger: VIGINUM changes its mandate or an oversight body identifies mischaracterization.

    Reopening trigger: Independent audits, litigation, parliamentary review, or affected-community evidence becomes available.

  13. CLAIM-REAL-13-PREDICTIVE Aggregate forecasting and person-level risk scoring are different system classesPRIMARY_OFFICIAL_MULTI_CASE_CLOSURE_PARTIAL
    Aggregate forecasting, case routing, family or person-level risk scoring, and institutional triage are different system classes. Current AFST, CARS, and FluSight records establish continued program activity, not universal validity, fairness, causal benefit, notice, appeal, or complete repair.

    Evidence supports

    Every predictive system should disclose unit, target, horizon, data, model role, threshold, action, evaluation, false-positive and false-negative burden, subgroup effects, human authority, feedback loops, appeal, and current status.

    Does not establish

    That all forecasting is harmful, that all person-level scoring is unlawful, or that a model’s operator description proves independent validity.

    Claim type
    PREDICTIVE_DEPLOYMENT_SYNTHESIS
    Scope
    Public health, humanitarian planning, conflict warning, policing, benefits, child welfare, and immigration workflow
    Source relationship
    Seven primary official records illustrate distinct deployments, authorities, units of prediction, and decision consequences.
    Currentness
    CURRENT_AS_OF_2026-07-29_WITH_STATED_LIMITS
    Specialist disposition
    NOT_COMPLETED

    Measurement limits

    Operator transparency records often omit independent calibration, subgroup error, and counterfactual outcomes.

    Generalization limits

    Health, humanitarian, conflict, visa, child-welfare, fraud, and policing systems have different purposes and legal bases.

    Privacy boundary

    Population planning should minimize reidentification; person-level systems require heightened data minimization and access control.

    Affected-person boundary

    No system should infer endorsement from affected communities; direct and representative evidence remains necessary.

    Competing explanations

    • Aggregate forecasting may support beneficial resource allocation without identifying individuals.
    • Human reviewers may mitigate or amplify model errors depending on authority and workload.

    Primary or authoritative records

    Correction trigger: An operator or court changes a system’s purpose, authority, status, or outcome.

    Reopening trigger: Independent evaluation, litigation, audit, or affected-person evidence becomes available.

THIRTEEN VERSIONED CORRECTION RECORDS

See what changed without erasing the prior wording

Every record preserves the stable claim ID, prior wording, current wording, changed fields, reason, affected public anchor, error status, and the evidence that should correct or reopen the record.

  1. 01Reversal is not complete repairCURRENTNESS_REFRESH
    Stable correction ID
    CORR-REAL-01-REPAIR
    Error status
    NO_ERROR_FOUND_CURRENTNESS_ONLY
    Effective date
    2026-07-28
    Editor/reviewer identity
    NOT_ATTRIBUTABLE_REPOSITORY_AGENT
    Specialist disposition
    NOT_COMPLETED

    Changed fields: public_wording, authority_ids, currentness_status

    Reason: Added 2026 remedy implementation context while retaining the incomplete-repair boundary.

    Prior wording

    A successful appeal, reversal, settlement, or restored account can stop an immediate error without repairing every copied record, derived score, ranking signal, vendor system, financial loss, reputational consequence, or psychological harm.

    Current wording

    A successful appeal, reversal, settlement, or restored account can stop an immediate error without repairing every copied record, derived score, ranking signal, vendor system, financial loss, reputational consequence, or psychological harm. The 2026 Robodebt settlement and resolution schemes document additional remedy stages, not proof of complete downstream repair.
    Field-by-field delta
    {
    "public_wording": {
        "prior": "A successful appeal, reversal, settlement, or restored account can stop an immediate error without repairing every copied record, derived score, ranking signal, vendor system, financial loss, reputational consequence, or psychological harm.",
        "current": "A successful appeal, reversal, settlement, or restored account can stop an immediate error without repairing every copied record, derived score, ranking signal, vendor system, financial loss, reputational consequence, or psychological harm. The 2026 Robodebt settlement and resolution schemes document additional remedy stages, not proof of complete downstream repair."
    },
    "authority_ids": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    },
    "currentness_status": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    }
}

    Affected public anchor: /psyops/ai-psyops/field-realism#claim-real-01-repair

    Correction trigger: A source shows complete downstream correction and remediation for a cited case.

    Reopening trigger: New audit, compensation, propagation, or repeat-error data becomes available.

  2. 02Deletion and output blocking are not model unlearningNARROWING
    Stable correction ID
    CORR-REAL-02-UNLEARNING
    Error status
    WORDING_REVISED_WITHOUT_CLAIM_OF_PRIOR_MISCONDUCT
    Effective date
    2026-07-28
    Editor/reviewer identity
    NOT_ATTRIBUTABLE_REPOSITORY_AGENT
    Specialist disposition
    NOT_COMPLETED

    Changed fields: public_wording, authority_ids, evidence_status

    Reason: New 2026 auditing research requires explicit separation of suppression, approximate unlearning, exact forgetting, and production validation.

    Prior wording

    Deleting a source record or blocking an output does not by itself remove the record’s learned influence from model weights, embeddings, features, memories, caches, backups, or downstream recipients.

    Current wording

    Source deletion, suppression, output blocking, model editing, retraining, and machine unlearning are distinct interventions. Current public audits do not establish that a named production foundation model has exactly removed a requested influence or reached retraining-equivalent behavior across hidden and observable pathways.
    Field-by-field delta
    {
    "public_wording": {
        "prior": "Deleting a source record or blocking an output does not by itself remove the record’s learned influence from model weights, embeddings, features, memories, caches, backups, or downstream recipients.",
        "current": "Source deletion, suppression, output blocking, model editing, retraining, and machine unlearning are distinct interventions. Current public audits do not establish that a named production foundation model has exactly removed a requested influence or reached retraining-equivalent behavior across hidden and observable pathways."
    },
    "authority_ids": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    },
    "evidence_status": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    }
}

    Affected public anchor: /psyops/ai-psyops/field-realism#claim-real-02-unlearning

    Correction trigger: NIST or a recognized standard adopts a materially different definition.

    Reopening trigger: Validated exact-unlearning benchmarks for large deployed models become available.

  3. 03Observed movement is not direct knowledge of emotionCLARIFICATION
    Stable correction ID
    CORR-REAL-03-MENTAL-STATE
    Error status
    WORDING_REVISED_WITHOUT_CLAIM_OF_PRIOR_MISCONDUCT
    Effective date
    2026-07-28
    Editor/reviewer identity
    NOT_ATTRIBUTABLE_REPOSITORY_AGENT
    Specialist disposition
    NOT_COMPLETED

    Changed fields: public_wording, currentness_status

    Reason: Restated the distinction between signal detection and named-construct validity after the refreshed field search.

    Prior wording

    A system may detect a facial movement, voice feature, gaze pattern, physiological signal, or interaction trace without validly determining a person’s emotion, deception, loyalty, attention, vulnerability, or intent.

    Current wording

    A system may detect a facial movement, vocal feature, gaze pattern, physiological signal, or interaction trace without reliably establishing the person’s emotion, deception, loyalty, vulnerability, intent, or dangerousness. Independent construct-specific field validity remains incomplete.
    Field-by-field delta
    {
    "public_wording": {
        "prior": "A system may detect a facial movement, voice feature, gaze pattern, physiological signal, or interaction trace without validly determining a person’s emotion, deception, loyalty, attention, vulnerability, or intent.",
        "current": "A system may detect a facial movement, vocal feature, gaze pattern, physiological signal, or interaction trace without reliably establishing the person’s emotion, deception, loyalty, vulnerability, intent, or dangerousness. Independent construct-specific field validity remains incomplete."
    },
    "currentness_status": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    }
}

    Affected public anchor: /psyops/ai-psyops/field-realism#claim-real-03-mental-state

    Correction trigger: Large replicated field studies establish robust construct validity for a specified system and use.

    Reopening trigger: New scientific consensus or regulatory guidance materially changes the boundary.

  4. 04Cognitive-liberty law is jurisdiction-specific and unevenNO_SUBSTANTIVE_CHANGE
    Stable correction ID
    CORR-REAL-04-LAW
    Error status
    NO_ERROR_FOUND_CURRENTNESS_ONLY
    Effective date
    2026-07-28
    Editor/reviewer identity
    NOT_ATTRIBUTABLE_REPOSITORY_AGENT
    Specialist disposition
    NOT_COMPLETED

    Changed fields: review_date

    Reason: Currentness was reviewed; no universal-law conclusion was added.

    Prior wording

    Current protections for neural data, emotion inference, manipulation, automated decisions, and mental privacy are distributed across jurisdictions and legal instruments; no single universal cognitive-liberty code governs every system.

    Current wording

    Cognitive-liberty, neural-data, biometric, automated-decision, and AI-manipulation rules remain jurisdiction-specific, staged, and uneven. A proposal, guidance document, settlement, trial judgment, or one jurisdiction’s enacted rule is not a universal code.
    Field-by-field delta
    {
    "review_date": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    }
}

    Affected public anchor: /psyops/ai-psyops/field-realism#claim-real-04-law

    Correction trigger: A law is amended, repealed, corrected, or authoritatively interpreted.

    Reopening trigger: Implementing rules, enforcement, or appellate decisions become available.

  5. 05Visibility actions and remedy outcomes are separateCLARIFICATION
    Stable correction ID
    CORR-REAL-05-VISIBILITY
    Error status
    WORDING_REVISED_WITHOUT_CLAIM_OF_PRIOR_MISCONDUCT
    Effective date
    2026-07-28
    Editor/reviewer identity
    NOT_ATTRIBUTABLE_REPOSITORY_AGENT
    Specialist disposition
    NOT_COMPLETED

    Changed fields: public_wording

    Reason: Made downstream ranking, recommendation, strike, and income repair explicit.

    Prior wording

    Removal, restriction, search exclusion, recommendation exclusion, downranking, labeling, monetization change, account penalties, synthesized reframing, and profile changes are different interventions with different notice, evidence, appeal, restoration, and downstream effects.

    Current wording

    Removal, restriction, search exclusion, recommendation exclusion, downranking, labeling, monetization change, account penalty, personalized refusal, synthesized reframing, and persistent-profile change are different actions. A visible restoration does not by itself show that ranking, recommendation, strikes, income, or downstream records were repaired.
    Field-by-field delta
    {
    "public_wording": {
        "prior": "Removal, restriction, search exclusion, recommendation exclusion, downranking, labeling, monetization change, account penalties, synthesized reframing, and profile changes are different interventions with different notice, evidence, appeal, restoration, and downstream effects.",
        "current": "Removal, restriction, search exclusion, recommendation exclusion, downranking, labeling, monetization change, account penalty, personalized refusal, synthesized reframing, and persistent-profile change are different actions. A visible restoration does not by itself show that ranking, recommendation, strikes, income, or downstream records were repaired."
    }
}

    Affected public anchor: /psyops/ai-psyops/field-realism#claim-real-05-visibility

    Correction trigger: Platform records show that a named action or outcome was misclassified.

    Reopening trigger: Audit data reveals downstream ranking, strike, monetization, or profile changes.

  6. 06Lived experience and institutional evidence require explicit scopeNO_SUBSTANTIVE_CHANGE
    Stable correction ID
    CORR-REAL-06-AFFECTED-EVIDENCE
    Error status
    NO_ERROR_FOUND_CURRENTNESS_ONLY
    Effective date
    2026-07-28
    Editor/reviewer identity
    NOT_ATTRIBUTABLE_REPOSITORY_AGENT
    Specialist disposition
    NOT_COMPLETED

    Changed fields: review_date

    Reason: Evidence-type and representativeness boundaries remain unchanged.

    Prior wording

    Attributed first-person, representative, institutional, aggregated, anonymized, and derivative evidence can document serious algorithmic harms, but none should be treated as automatically representative of every affected person or as proof of one exclusive causal mechanism.

    Current wording

    First-person, representative, institutional, aggregated, anonymized, and derivative evidence can document different parts of an AI-governance harm. None should be silently converted into prevalence, universal causation, representativeness, or affected-community endorsement.
    Field-by-field delta
    {
    "review_date": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    }
}

    Affected public anchor: /psyops/ai-psyops/field-realism#claim-real-06-affected-evidence

    Correction trigger: A cited account is corrected, withdrawn, or shown to be misrepresented.

    Reopening trigger: Authorized community research, representative surveys, or new official records become available.

  7. 07Regulatory inquiry is not clinical causalitySOURCE_ADDITION
    Stable correction ID
    CORR-REAL-07-COMPANION-SAFETY
    Error status
    WORDING_REVISED_WITHOUT_CLAIM_OF_PRIOR_MISCONDUCT
    Effective date
    2026-07-28
    Editor/reviewer identity
    NOT_ATTRIBUTABLE_REPOSITORY_AGENT
    Specialist disposition
    NOT_COMPLETED

    Changed fields: public_wording, authority_ids, currentness_status

    Reason: Added April 2026 FTC testimony confirming the study remained ongoing without public findings.

    Prior wording

    The FTC’s companion-chatbot inquiry establishes an official demand for information about safety, children and teens, disclosures, monetization, and data practices; it is not a completed finding that any product caused a psychiatric condition.

    Current wording

    The FTC’s companion-chatbot work remained publicly described as an ongoing Section 6(b) study in April 2026. An inquiry documents regulatory concern and information gathering; it is not a completed finding of wrongdoing, clinical causation, or prevalence.
    Field-by-field delta
    {
    "public_wording": {
        "prior": "The FTC’s companion-chatbot inquiry establishes an official demand for information about safety, children and teens, disclosures, monetization, and data practices; it is not a completed finding that any product caused a psychiatric condition.",
        "current": "The FTC’s companion-chatbot work remained publicly described as an ongoing Section 6(b) study in April 2026. An inquiry documents regulatory concern and information gathering; it is not a completed finding of wrongdoing, clinical causation, or prevalence."
    },
    "authority_ids": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    },
    "currentness_status": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    }
}

    Affected public anchor: /psyops/ai-psyops/field-realism#claim-real-07-companion-safety

    Correction trigger: FTC publishes findings or corrects the inquiry scope.

    Reopening trigger: Peer-reviewed longitudinal evidence, final enforcement, or adjudicated product records become available.

  8. 08An incident record does not automatically prove behavioral effectNO_SUBSTANTIVE_CHANGE
    Stable correction ID
    CORR-REAL-08-INCIDENT-EFFECT
    Error status
    NO_ERROR_FOUND_CURRENTNESS_ONLY
    Effective date
    2026-07-28
    Editor/reviewer identity
    NOT_ATTRIBUTABLE_REPOSITORY_AGENT
    Specialist disposition
    NOT_COMPLETED

    Changed fields: review_date

    Reason: The complete effect chain remains the controlling interpretation.

    Prior wording

    The New Hampshire Biden robocall is documented as an AI-generated impersonation, spoofed distribution event, and enforcement matter; the available authority does not establish how many recipients believed it or whether voting behavior changed.

    Current wording

    A verified artifact, coordinated distribution, operator record, or enforcement action can establish an incident without proving recipient exposure, belief, intention, behavior change, operational outcome, or strategic effect.
    Field-by-field delta
    {
    "review_date": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    }
}

    Affected public anchor: /psyops/ai-psyops/field-realism#claim-real-08-incident-effect

    Correction trigger: FCC or a court changes the operator, scope, or final disposition.

    Reopening trigger: Survey, telecom, or election research measures exposure, belief, or behavior.

  9. 09Behavioral coordination evidence is stronger than a synthetic-content score aloneCLARIFICATION
    Stable correction ID
    CORR-REAL-09-SWARM-DETECTION
    Error status
    WORDING_REVISED_WITHOUT_CLAIM_OF_PRIOR_MISCONDUCT
    Effective date
    2026-07-28
    Editor/reviewer identity
    NOT_ATTRIBUTABLE_REPOSITORY_AGENT
    Specialist disposition
    NOT_COMPLETED

    Changed fields: public_wording

    Reason: Separated coordination evidence from synthetic-origin and sponsor attribution.

    Prior wording

    Synthetic profile imagery or an AI-content detector score can be one clue, but coordinated inauthentic behavior should be assessed through network, chronology, infrastructure, role, and platform-enforcement evidence; a generated face alone does not identify a sponsor or prove effect.

    Current wording

    Behavioral and infrastructure evidence can establish coordination more strongly than an AI-content score alone. Coordination still does not automatically establish actor identity, sponsorship, intent, audience effect, or the synthetic origin of every asset.
    Field-by-field delta
    {
    "public_wording": {
        "prior": "Synthetic profile imagery or an AI-content detector score can be one clue, but coordinated inauthentic behavior should be assessed through network, chronology, infrastructure, role, and platform-enforcement evidence; a generated face alone does not identify a sponsor or prove effect.",
        "current": "Behavioral and infrastructure evidence can establish coordination more strongly than an AI-content score alone. Coordination still does not automatically establish actor identity, sponsorship, intent, audience effect, or the synthetic origin of every asset."
    }
}

    Affected public anchor: /psyops/ai-psyops/field-realism#claim-real-09-swarm-detection

    Correction trigger: A platform retracts or materially changes a network attribution.

    Reopening trigger: Independent forensic, legal, or financial evidence establishes actor or sponsor identity.

  10. 10Official debunking closes some stages, not all effectsNARROWING
    Stable correction ID
    CORR-REAL-10-DEEPFAKE-RESPONSE
    Error status
    WORDING_REVISED_WITHOUT_CLAIM_OF_PRIOR_MISCONDUCT
    Effective date
    2026-07-28
    Editor/reviewer identity
    NOT_ATTRIBUTABLE_REPOSITORY_AGENT
    Specialist disposition
    NOT_COMPLETED

    Changed fields: public_wording, evidence_status

    Reason: Removed any implication that ridicule or rapid debunking established recipient belief or operational effect.

    Prior wording

    Official Ukrainian records document the Zelenskyy surrender deepfake and later Zaluzhnyi deepfakes and describe rapid detection and debunking; they do not quantify continuing belief, troop-level exposure, operator identity for every artifact, or strategic effect.

    Current wording

    The Zelenskyy surrender deepfake and rapid rebuttal establish the artifact, hacked distribution, and authentication response. Public sources reviewed here do not establish representative recipient belief, frontline behavior change, operational outcome, or strategic effect.
    Field-by-field delta
    {
    "public_wording": {
        "prior": "Official Ukrainian records document the Zelenskyy surrender deepfake and later Zaluzhnyi deepfakes and describe rapid detection and debunking; they do not quantify continuing belief, troop-level exposure, operator identity for every artifact, or strategic effect.",
        "current": "The Zelenskyy surrender deepfake and rapid rebuttal establish the artifact, hacked distribution, and authentication response. Public sources reviewed here do not establish representative recipient belief, frontline behavior change, operational outcome, or strategic effect."
    },
    "evidence_status": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    }
}

    Affected public anchor: /psyops/ai-psyops/field-realism#claim-real-10-deepfake-response

    Correction trigger: The issuing body or stronger forensic evidence changes chronology or attribution.

    Reopening trigger: Platform telemetry, intelligence release, or recipient research measures exposure and effect.

  11. 11Provenance is not truth, and the liar’s dividend is context-dependentCLARIFICATION
    Stable correction ID
    CORR-REAL-11-LIARS-DIVIDEND
    Error status
    WORDING_REVISED_WITHOUT_CLAIM_OF_PRIOR_MISCONDUCT
    Effective date
    2026-07-28
    Editor/reviewer identity
    NOT_ATTRIBUTABLE_REPOSITORY_AGENT
    Specialist disposition
    NOT_COMPLETED

    Changed fields: public_wording, authority_ids

    Reason: Integrated the standard’s own statement that embedded metadata can be removed and that soft bindings aid recovery rather than guarantee survival.

    Prior wording

    Cryptographic provenance can verify a signed history or claim, but it does not prove the represented event is true, and missing credentials do not prove media is false. Experimental evidence also shows that false deepfake claims do not work uniformly across formats and contexts.

    Current wording

    C2PA can authenticate signed provenance claims, and durable credentials can use watermarking or fingerprinting to recover separated manifests. Embedded metadata can still be removed; credentials do not prove the depicted event is true, and their absence does not make media false. Liar’s-dividend effects remain context-dependent.
    Field-by-field delta
    {
    "public_wording": {
        "prior": "Cryptographic provenance can verify a signed history or claim, but it does not prove the represented event is true, and missing credentials do not prove media is false. Experimental evidence also shows that false deepfake claims do not work uniformly across formats and contexts.",
        "current": "C2PA can authenticate signed provenance claims, and durable credentials can use watermarking or fingerprinting to recover separated manifests. Embedded metadata can still be removed; credentials do not prove the depicted event is true, and their absence does not make media false. Liar’s-dividend effects remain context-dependent."
    },
    "authority_ids": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    }
}

    Affected public anchor: /psyops/ai-psyops/field-realism#claim-real-11-liars-dividend

    Correction trigger: C2PA or the journal corrects the relevant records.

    Reopening trigger: Field evidence on credential adoption, attack resistance, or liar’s-dividend effects becomes available.

  12. 12Democratic defense should target behavior and foreign interference, not lawful beliefCURRENTNESS_REFRESH
    Stable correction ID
    CORR-REAL-12-DEMOCRATIC-DEFENSE
    Error status
    NO_ERROR_FOUND_CURRENTNESS_ONLY
    Effective date
    2026-07-28
    Editor/reviewer identity
    NOT_ATTRIBUTABLE_REPOSITORY_AGENT
    Specialist disposition
    NOT_COMPLETED

    Changed fields: public_wording, authority_ids, currentness_status

    Reason: Added the 2026 VIGINUM remit update while preserving false-positive, remedy, and affected-community gaps.

    Prior wording

    A defensible cognitive-security mandate can focus on concealed foreign direction, artificial amplification, impersonation, technical tampering, and coordinated inauthentic behavior while preserving lawful domestic belief, dissent, anonymity, journalism, and political opposition.

    Current wording

    Democratic cognitive defense should target concealed coordination, foreign direction, impersonation, technical tampering, artificial amplification, fraud, and platform abuse rather than lawful belief. VIGINUM’s 2026 remit update documents institutional currentness, not an independently verified false-positive or remedy record.
    Field-by-field delta
    {
    "public_wording": {
        "prior": "A defensible cognitive-security mandate can focus on concealed foreign direction, artificial amplification, impersonation, technical tampering, and coordinated inauthentic behavior while preserving lawful domestic belief, dissent, anonymity, journalism, and political opposition.",
        "current": "Democratic cognitive defense should target concealed coordination, foreign direction, impersonation, technical tampering, artificial amplification, fraud, and platform abuse rather than lawful belief. VIGINUM’s 2026 remit update documents institutional currentness, not an independently verified false-positive or remedy record."
    },
    "authority_ids": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    },
    "currentness_status": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    }
}

    Affected public anchor: /psyops/ai-psyops/field-realism#claim-real-12-democratic-defense

    Correction trigger: VIGINUM changes its mandate or an oversight body identifies mischaracterization.

    Reopening trigger: Independent audits, litigation, parliamentary review, or affected-community evidence becomes available.

  13. 13Aggregate forecasting and person-level risk scoring are different system classesCURRENTNESS_REFRESH
    Stable correction ID
    CORR-REAL-13-PREDICTIVE
    Error status
    NO_ERROR_FOUND_CURRENTNESS_ONLY
    Effective date
    2026-07-28
    Editor/reviewer identity
    NOT_ATTRIBUTABLE_REPOSITORY_AGENT
    Specialist disposition
    NOT_COMPLETED

    Changed fields: public_wording, authority_ids, currentness_status

    Reason: Added current AFST, CARS, and FluSight program-status records without upgrading validity claims.

    Prior wording

    Population-level forecasts used for planning must be separated from person- or family-level risk scores used to route, investigate, screen, deny, or intensify intervention. Accuracy, authority, rights burden, feedback loops, and remedy differ substantially between these system classes.

    Current wording

    Aggregate forecasting, case routing, family or person-level risk scoring, and institutional triage are different system classes. Current AFST, CARS, and FluSight records establish continued program activity, not universal validity, fairness, causal benefit, notice, appeal, or complete repair.
    Field-by-field delta
    {
    "public_wording": {
        "prior": "Population-level forecasts used for planning must be separated from person- or family-level risk scores used to route, investigate, screen, deny, or intensify intervention. Accuracy, authority, rights burden, feedback loops, and remedy differ substantially between these system classes.",
        "current": "Aggregate forecasting, case routing, family or person-level risk scoring, and institutional triage are different system classes. Current AFST, CARS, and FluSight records establish continued program activity, not universal validity, fairness, causal benefit, notice, appeal, or complete repair."
    },
    "authority_ids": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    },
    "currentness_status": {
        "prior": "WIP57_VALUE_RETAINED_OR_VERSIONED",
        "current": "WIP58_VERSIONED_VALUE"
    }
}

    Affected public anchor: /psyops/ai-psyops/field-realism#claim-real-13-predictive

    Correction trigger: An operator or court changes a system’s purpose, authority, status, or outcome.

    Reopening trigger: Independent evaluation, litigation, audit, or affected-person evidence becomes available.

THIRTEEN CONTRADICTION AND ORIGIN-CHAIN RECORDS

Preserve agreement, conflict, edition differences, and unresolved scope

Two sources can agree on an institutional event while differing on validity, effect, currentness, remedy, or independence. The selected public wording is shown without silently combining incompatible propositions.

  1. 01A reversal or compensation scheme is complete repair.UNRESOLVED_OUTSIDE_PUBLIC_SOURCE_SCOPE
    Stable ID
    CONTRA-01-REPAIR
    Source relationship
    INDEPENDENT_RECORDS_FROM_SAME_RESPONSIBLE_INSTITUTIONAL_ECOSYSTEM
    Dimensions affected
    content_status, currentness, effect, remedy, source_independence

    Agreement

    Both records document bounded remedy stages.

    Conflict or unresolved difference

    Neither record establishes complete correction of all source, inferred, copied, or downstream data or all unresolved harm.

    Date or edition difference: Versioned records are preserved; no silent overwrite or harmonization.

    Public wording selected: Describe documented compensation and resolution stages while keeping complete downstream repair unresolved.

    Prohibited harmonization: Do not collapse source scope, procedural status, validity, attribution, effect, or remedy into one stronger conclusion.

    Compared records

    Correction trigger: A primary or independently attributable record resolves the identified conflict or corrects a source.

    Reopening trigger: Reopen when a new edition, adjudication, audit, field study, affected-person record, or downstream-repair record appears.

  2. 02An unlearning label proves exact forgetting.RESOLVABLE_BY_SCOPE
    Stable ID
    CONTRA-02-UNLEARNING
    Source relationship
    INDEPENDENT_DEFINITION_AND_RESEARCH_AUDIT
    Dimensions affected
    content_status, currentness, effect, remedy, source_independence

    Agreement

    Both recognize a goal of removing specified influence.

    Conflict or unresolved difference

    The audit literature finds that reliable proof of true forgetting remains open and that ordinary tests can overstate success.

    Date or edition difference: Versioned records are preserved; no silent overwrite or harmonization.

    Public wording selected: Use “unlearning” for the claimed intervention and separately state exactness, audit method, and production evidence.

    Prohibited harmonization: Do not collapse source scope, procedural status, validity, attribution, effect, or remedy into one stronger conclusion.

    Compared records

    Correction trigger: A primary or independently attributable record resolves the identified conflict or corrects a source.

    Reopening trigger: Reopen when a new edition, adjudication, audit, field study, affected-person record, or downstream-repair record appears.

  3. 03Accurate signal detection establishes an inner mental state.RESOLVABLE_BY_MEASUREMENT_LAYER
    Stable ID
    CONTRA-03-MENTAL-STATE
    Source relationship
    INDEPENDENT_APPLIED_AND_FOUNDATIONAL_RESEARCH
    Dimensions affected
    content_status, currentness, effect, remedy, source_independence

    Agreement

    Both concern systems that map observable behavior to emotional labels.

    Conflict or unresolved difference

    The foundational review disputes a universal one-to-one mapping from facial movement to emotion, while deployment records may report classifier performance on selected labels.

    Date or edition difference: Versioned records are preserved; no silent overwrite or harmonization.

    Public wording selected: Report sensor, feature, label, construct, calibration, context, subgroup, and institutional-action findings separately.

    Prohibited harmonization: Do not collapse source scope, procedural status, validity, attribution, effect, or remedy into one stronger conclusion.

    Compared records

    Correction trigger: A primary or independently attributable record resolves the identified conflict or corrects a source.

    Reopening trigger: Reopen when a new edition, adjudication, audit, field study, affected-person record, or downstream-repair record appears.

  4. 04Cognitive liberty is already one universal enacted legal code.RESOLVED_AS_JURISDICTION_SPECIFIC
    Stable ID
    CONTRA-04-LAW
    Source relationship
    INDEPENDENT_JURISDICTION_SPECIFIC_LEGAL_RECORDS
    Dimensions affected
    content_status, currentness, effect, remedy, source_independence

    Agreement

    Both regulate bounded AI or biological-data practices.

    Conflict or unresolved difference

    They use different authorities, scopes, effective dates, remedies, and jurisdictions.

    Date or edition difference: Versioned records are preserved; no silent overwrite or harmonization.

    Public wording selected: Describe each legal instrument by jurisdiction and lifecycle; do not synthesize universal law.

    Prohibited harmonization: Do not collapse source scope, procedural status, validity, attribution, effect, or remedy into one stronger conclusion.

    Compared records

    Correction trigger: A primary or independently attributable record resolves the identified conflict or corrects a source.

    Reopening trigger: Reopen when a new edition, adjudication, audit, field study, affected-person record, or downstream-repair record appears.

  5. 05Restoration of a post proves all visibility and account effects were repaired.UNRESOLVED_MISSING_DOWNSTREAM_EVIDENCE
    Stable ID
    CONTRA-05-VISIBILITY
    Source relationship
    ONE_ORIGIN_CHAIN_WITH_MISSING_DOWNSTREAM_RECORD
    Dimensions affected
    content_status, currentness, effect, remedy, source_independence

    Agreement

    The decision establishes reversal and restoration.

    Conflict or unresolved difference

    The same decision does not disclose prior-reach restoration, recommendation eligibility, strike propagation, income effects, or copied internal signals.

    Date or edition difference: Versioned records are preserved; no silent overwrite or harmonization.

    Public wording selected: State restoration only; list downstream repair as NOT_RETRIEVED.

    Prohibited harmonization: Do not collapse source scope, procedural status, validity, attribution, effect, or remedy into one stronger conclusion.

    Compared records

    Correction trigger: A primary or independently attributable record resolves the identified conflict or corrects a source.

    Reopening trigger: Reopen when a new edition, adjudication, audit, field study, affected-person record, or downstream-repair record appears.

  6. 06A documented affected-person account establishes population prevalence.RESOLVED_BY_EVIDENCE_TYPE
    Stable ID
    CONTRA-06-AFFECTED-EVIDENCE
    Source relationship
    INDEPENDENT_INQUIRY_AND_REMEDY_RECORDS_WITH_SHARED_EVENT
    Dimensions affected
    content_status, currentness, effect, remedy, source_independence

    Agreement

    Both document serious institutional harm and affected groups.

    Conflict or unresolved difference

    Neither converts selected testimony or settlement membership into a universal prevalence estimate for algorithmic harms.

    Date or edition difference: Versioned records are preserved; no silent overwrite or harmonization.

    Public wording selected: Label testimony, representative, aggregate, and institutional evidence separately.

    Prohibited harmonization: Do not collapse source scope, procedural status, validity, attribution, effect, or remedy into one stronger conclusion.

    Compared records

    Correction trigger: A primary or independently attributable record resolves the identified conflict or corrects a source.

    Reopening trigger: Reopen when a new edition, adjudication, audit, field study, affected-person record, or downstream-repair record appears.

  7. 07The FTC inquiry is a finding that companion chatbots caused clinical harm.RESOLVED_BY_PROCEDURAL_STATUS
    Stable ID
    CONTRA-07-COMPANION
    Source relationship
    DERIVATIVE_AGENCY_CURRENTNESS_CHAIN
    Dimensions affected
    content_status, currentness, effect, remedy, source_independence

    Agreement

    Both confirm the study and its safety questions.

    Conflict or unresolved difference

    Neither is a product-specific liability finding, clinical diagnosis, prevalence study, or causal determination.

    Date or edition difference: Versioned records are preserved; no silent overwrite or harmonization.

    Public wording selected: Use “inquiry” or “study”; do not use “finding” or “enforcement” unless later records support it.

    Prohibited harmonization: Do not collapse source scope, procedural status, validity, attribution, effect, or remedy into one stronger conclusion.

    Compared records

    Correction trigger: A primary or independently attributable record resolves the identified conflict or corrects a source.

    Reopening trigger: Reopen when a new edition, adjudication, audit, field study, affected-person record, or downstream-repair record appears.

  8. 08A forfeiture order proves voters changed behavior.RESOLVED_BY_EFFECT_STAGE
    Stable ID
    CONTRA-08-ROBOCALL
    Source relationship
    DERIVATIVE_AGENCY_ENFORCEMENT_CHAIN
    Dimensions affected
    content_status, currentness, effect, remedy, source_independence

    Agreement

    Both establish the robocall action and agency disposition.

    Conflict or unresolved difference

    Neither measures belief, turnout, vote choice, or strategic election effect.

    Date or edition difference: Versioned records are preserved; no silent overwrite or harmonization.

    Public wording selected: Report artifact, operator, distribution, and enforcement separately from behavioral effect.

    Prohibited harmonization: Do not collapse source scope, procedural status, validity, attribution, effect, or remedy into one stronger conclusion.

    Compared records

    Correction trigger: A primary or independently attributable record resolves the identified conflict or corrects a source.

    Reopening trigger: Reopen when a new edition, adjudication, audit, field study, affected-person record, or downstream-repair record appears.

  9. 09AI-generated-looking text proves a coordinated state operation.RESOLVABLE_BY_ATTRIBUTION_LAYER
    Stable ID
    CONTRA-09-SWARM
    Source relationship
    INDEPENDENT_PLATFORM_AND_PROVIDER_REPORTS_WITH_POSSIBLE_SHARED_INCIDENTS
    Dimensions affected
    content_status, currentness, effect, remedy, source_independence

    Agreement

    Both document coordinated deceptive activity and platform/provider disruption.

    Conflict or unresolved difference

    Content-generation indicators do not alone establish coordination, and coordination does not automatically establish sponsor, intent, or effect.

    Date or edition difference: Versioned records are preserved; no silent overwrite or harmonization.

    Public wording selected: Prioritize behavior, infrastructure, and cross-source evidence; keep sponsor and effect separately graded.

    Prohibited harmonization: Do not collapse source scope, procedural status, validity, attribution, effect, or remedy into one stronger conclusion.

    Compared records

    Correction trigger: A primary or independently attributable record resolves the identified conflict or corrects a source.

    Reopening trigger: Reopen when a new edition, adjudication, audit, field study, affected-person record, or downstream-repair record appears.

  10. 10Rapid ridicule and debunking prove no operational effect.UNRESOLVED_RECIPIENT_AND_OPERATIONAL_EVIDENCE
    Stable ID
    CONTRA-10-UKRAINE
    Source relationship
    INDEPENDENT_INCIDENT_ANALYSIS_AND_CONTROLLED_RESEARCH
    Dimensions affected
    content_status, currentness, effect, remedy, source_independence

    Agreement

    Both support concern about the artifact and trust consequences.

    Conflict or unresolved difference

    Incident reporting does not supply representative frontline exposure or behavior data, and controlled research cannot retroactively measure that incident’s operational effect.

    Date or edition difference: Versioned records are preserved; no silent overwrite or harmonization.

    Public wording selected: State the rapid response and public reaction without claiming zero or positive strategic effect.

    Prohibited harmonization: Do not collapse source scope, procedural status, validity, attribution, effect, or remedy into one stronger conclusion.

    Compared records

    Correction trigger: A primary or independently attributable record resolves the identified conflict or corrects a source.

    Reopening trigger: Reopen when a new edition, adjudication, audit, field study, affected-person record, or downstream-repair record appears.

  11. 11A valid Content Credential proves the depicted event occurred and persists through every transformation.RESOLVED_BY_STANDARD_SCOPE
    Stable ID
    CONTRA-11-C2PA
    Source relationship
    SAME_STANDARD_ORIGIN_CHAIN
    Dimensions affected
    content_status, currentness, effect, remedy, source_independence

    Agreement

    Both establish a signed provenance architecture and durable-credential mechanisms.

    Conflict or unresolved difference

    The explainer states metadata can be removed; the standard authenticates claims about provenance, not real-world truth of the depicted event.

    Date or edition difference: Versioned records are preserved; no silent overwrite or harmonization.

    Public wording selected: Describe credential validation, removal, recovery, and truth limits separately.

    Prohibited harmonization: Do not collapse source scope, procedural status, validity, attribution, effect, or remedy into one stronger conclusion.

    Compared records

    Correction trigger: A primary or independently attributable record resolves the identified conflict or corrects a source.

    Reopening trigger: Reopen when a new edition, adjudication, audit, field study, affected-person record, or downstream-repair record appears.

  12. 12Official operation counts establish independent accuracy and low false-positive burden.UNRESOLVED_PUBLIC_ACCOUNTABILITY_EVIDENCE
    Stable ID
    CONTRA-12-VIGINUM
    Source relationship
    OPERATOR_AND_OVERSIGHT_RECORDS_NOT_FULLY_INDEPENDENT_OF_INSTITUTION
    Dimensions affected
    content_status, currentness, effect, remedy, source_independence

    Agreement

    Both document institutional activity and an ethics/oversight layer.

    Conflict or unresolved difference

    Public records reviewed do not provide a complete false-positive register, affected-community study, appeal outcomes, or correction/remedy metrics.

    Date or edition difference: Versioned records are preserved; no silent overwrite or harmonization.

    Public wording selected: State operator-reported detections and oversight architecture; do not infer independent accuracy or absence of harm.

    Prohibited harmonization: Do not collapse source scope, procedural status, validity, attribution, effect, or remedy into one stronger conclusion.

    Compared records

    Correction trigger: A primary or independently attributable record resolves the identified conflict or corrects a source.

    Reopening trigger: Reopen when a new edition, adjudication, audit, field study, affected-person record, or downstream-repair record appears.

  13. 13An active operator-described predictive system is independently validated and beneficial.PARTIALLY_RESOLVABLE_FIELDS_REMAIN_OPEN
    Stable ID
    CONTRA-13-PREDICTIVE
    Source relationship
    OPERATOR_CURRENTNESS_AND_EVALUATION_RECORDS
    Dimensions affected
    content_status, currentness, effect, remedy, source_independence

    Agreement

    Both establish that AFST exists and has been evaluated.

    Conflict or unresolved difference

    Program currentness, selected evaluation findings, calibration, subgroup burden, causal child outcomes, notice, and remedy are different propositions.

    Date or edition difference: Versioned records are preserved; no silent overwrite or harmonization.

    Public wording selected: Use active status only for program currentness and preserve separate validity, fairness, outcome, and remedy findings.

    Prohibited harmonization: Do not collapse source scope, procedural status, validity, attribution, effect, or remedy into one stronger conclusion.

    Compared records

    Correction trigger: A primary or independently attributable record resolves the identified conflict or corrects a source.

    Reopening trigger: Reopen when a new edition, adjudication, audit, field study, affected-person record, or downstream-repair record appears.

FIFTY-SIX DIMENSIONAL SOURCE REVIEWS

Quality and independence without an opaque score

A primary institutional record may be authoritative for its own action while remaining non-independent for fairness, validity, effect, or remedy. Each dimension stays visible rather than being compressed into a single numerical grade.

Source relationship, independence, methods transparency, data availability, sample disclosure, currentness, and affected-person evidence
SourceRelationshipIndependenceMethodsDataSample disclosureCurrentnessAffected-person evidence
FTC inquiry into AI chatbots acting as companionsAUTH-01-FTC-COMPANION-INQUIRYPRIMARY_ISSUING_BODY_NOTICENOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
FCC final action concerning AI-generated Biden robocallsAUTH-02-FCC-BIDEN-ROBOCALL-FINEPRIMARY_ISSUING_BODY_FINAL_ACTIONNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingFINAL_AGENCY_ACTION_AS_RECORDEDNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Meta Oversight Board decision: Ayahuasca brewAUTH-03-META-OVERSIGHT-AYAHUASCAPRIMARY_APPELLATE_DECISIONPARTIAL_OR_UNRESOLVEDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingDECISION_FINAL_WITHIN_BOARD_PROCESSNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Report of the Royal Commission into the Robodebt SchemeAUTH-04-ROBODEBT-ROYAL-COMMISSIONPRIMARY_OFFICIAL_INQUIRYNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingFINAL_REPORT_AS_PUBLISHEDNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Regulation (EU) 2024/1689 — Artificial Intelligence ActAUTH-05-EU-AI-ACTPRIMARY_LEGAL_TEXTPARTIAL_OR_UNRESOLVEDPARTIAL_PUBLIC_METHODSLIMITEDNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingENACTED_WITH_PHASED_APPLICATIONNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Colorado HB24-1058 — Protect Privacy of Biological DataAUTH-06-COLORADO-NEURAL-DATAPRIMARY_LEGISLATIVE_RECORDPARTIAL_OR_UNRESOLVEDPARTIAL_PUBLIC_METHODSLIMITEDNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingENACTED_STATE_LAW_AS_RECORDEDNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
C2PA Technical Specification 2.4AUTH-07-C2PA-SPEC-2-4PRIMARY_STANDARD_TEXTPARTIAL_OR_UNRESOLVEDHIGH_FOR_STATED_PROCEDURELIMITEDNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingVERSION_2_4_AS_ACCESSEDNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
NIST glossary entry: Machine UnlearningAUTH-08-NIST-MACHINE-UNLEARNINGPRIMARY_OFFICIAL_DEFINITIONNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDHIGH_FOR_STATED_PROCEDUREPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_ACCESSEDNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Machine Unlearning (SISA training architecture)AUTH-09-SISA-UNLEARNINGPRIMARY_RESEARCH_ARTICLEPARTIAL_OR_UNRESOLVEDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingHISTORICAL_FOUNDATIONAL_RESEARCHNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Emotional Expressions Reconsidered: Challenges to Inferring Emotion From Human Facial MovementsAUTH-10-BARRETT-FACIAL-MOVEMENTSPRIMARY_PEER_REVIEWED_REVIEWINDEPENDENT_FOR_LIMITED_PROPOSITIONPARTIAL_PUBLIC_METHODSLIMITEDNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingFOUNDATIONAL_REVIEW_STILL_RELEVANTNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Advisory Concerning the Chicago Police Department’s Predictive Risk ModelsAUTH-11-CHICAGO-PREDICTIVE-RISKPRIMARY_OVERSIGHT_RECORDINDEPENDENT_FOR_LIMITED_PROPOSITIONPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingHISTORICAL_DEPLOYMENT_DECOMMISSIONEDNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Allegheny Family Screening ToolAUTH-12-ALLEGHENY-FAMILY-SCREENINGPRIMARY_OPERATOR_DESCRIPTIONNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_OPERATOR_DESCRIPTION_AS_ACCESSEDNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
District Court of The Hague judgment concerning SyRIAUTH-13-SYRI-JUDGMENTPRIMARY_JUDICIAL_RECORDPARTIAL_OR_UNRESOLVEDPARTIAL_PUBLIC_METHODSLIMITEDNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingJUDGMENT_AS_PUBLISHEDNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
CDC FluSight forecastingAUTH-14-CDC-FLUSIGHTPRIMARY_PROGRAM_DESCRIPTIONNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_PROGRAM_AS_ACCESSEDNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Home Office Complexity Application Routing Solution — VisitsAUTH-15-HOME-OFFICE-CARSPRIMARY_OPERATOR_TRANSPARENCY_RECORDNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_RECORD_ACCESSEDNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
UNHCR predictive analytics for forced displacementAUTH-16-UNHCR-DISPLACEMENT-FORECASTPRIMARY_INSTITUTIONAL_TECHNICAL_RECORDNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDHIGH_FOR_STATED_PROCEDUREPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizing2025_TECHNICAL_RECORDNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Dynamic Conflict Risk Model technical reportAUTH-17-EU-JRC-DYNAMIC-CONFLICT-RISKPRIMARY_OFFICIAL_RESEARCHNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizing2025_REPORT_AS_PUBLISHEDNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Meta 2022 coordinated inauthentic behavior enforcement reportAUTH-18-META-CIB-2022PRIMARY_PLATFORM_ENFORCEMENT_REPORTPARTIAL_OR_UNRESOLVEDPARTIAL_PUBLIC_METHODSLIMITEDNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingHISTORICAL_2022_REPORTNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Disrupting deceptive uses of AI by covert influence operationsAUTH-19-OPENAI-COVERT-IOPRIMARY_PROVIDER_REPORTPARTIAL_OR_UNRESOLVEDPARTIAL_PUBLIC_METHODSLIMITEDNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingHISTORICAL_2024_REPORTNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
The Liar’s Dividend: Can Politicians Claim Misinformation to Evade Accountability?AUTH-20-LIARS-DIVIDEND-APSRPRIMARY_PEER_REVIEWED_EXPERIMENTSINDEPENDENT_FOR_LIMITED_PROPOSITIONPARTIAL_PUBLIC_METHODSLIMITEDNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingPUBLISHED_EXPERIMENTAL_EVIDENCENOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Information threat linked to artificial intelligenceAUTH-21-VIGINUM-AI-THREATPRIMARY_ISSUING_BODY_REPORTNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizing2025_REPORT_AS_PUBLISHEDNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Analytical report on AI-generated videos used to discredit Ukraine’s Defence ForcesAUTH-22-UKRAINE-CPD-DEEPFAKESPRIMARY_TARGETED_STATE_REPORTNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingHISTORICAL_INCIDENT_REPORTNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Predictions Put Into Practice: A Quasi-Experimental Evaluation of Chicago’s Predictive Policing PilotAUTH-23-CHICAGO-NIJ-EVALUATIONOFFICIAL_RESEARCH_SUMMARY_OF_QUASI_EXPERIMENTAL_EVALUATIONNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Parliamentary answers concerning SyRI data destruction and decision useAUTH-24-SYRI-DATA-DESTRUCTIONPRIMARY_EXECUTIVE_ANSWER_TO_PARLIAMENTNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Impact Evaluation of the Allegheny Family Screening Tool — Phase 2AUTH-25-AFST-PHASE2-EVALUATIONOPERATOR_COMMISSIONED_MULTIVARIABLE_EVALUATIONNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDHIGH_FOR_STATED_PROCEDUREPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Algorithms, Humans and Racial Disparities in Child Protection SystemsAUTH-26-AFST-HUMAN-RACIAL-DISPARITYACADEMIC_EMPIRICAL_ANALYSISINDEPENDENT_FOR_LIMITED_PROPOSITIONHIGH_FOR_STATED_PROCEDUREPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
An inspection of visit visa operations, December 2022 to January 2023AUTH-27-CARS-ICIBI-INSPECTIONINDEPENDENT_INSPECTION_OF_OPERATOR_WORKFLOWINDEPENDENT_FOR_LIMITED_PROPOSITIONHIGH_FOR_STATED_PROCEDUREPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Home Office response to inspection of visit visa operationsAUTH-28-CARS-HOME-OFFICE-RESPONSEPRIMARY_OPERATOR_RESPONSE_TO_INDEPENDENT_INSPECTIONINDEPENDENT_FOR_LIMITED_PROPOSITIONPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
FluSight 2024–2025 Season EvaluationAUTH-29-FLUSIGHT-2024-25-EVALUATIONPRIMARY_OPERATOR_PERFORMANCE_EVALUATIONNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDHIGH_FOR_STATED_PROCEDUREPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
FluSight 2023–2024 Season EvaluationAUTH-30-FLUSIGHT-2023-24-EVALUATIONPRIMARY_OPERATOR_PERFORMANCE_EVALUATIONNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDHIGH_FOR_STATED_PROCEDUREPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Project Jetson technical specificationsAUTH-31-UNHCR-JETSON-TECHNICALPRIMARY_OPERATOR_TECHNICAL_DESCRIPTIONNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDHIGH_FOR_STATED_PROCEDUREPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingHISTORICAL_TECHNICAL_RECORD_CURRENTNESS_BOUNDEDNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
UNHCR AI ApproachAUTH-32-UNHCR-AI-APPROACH-2025PRIMARY_OPERATOR_POLICY_DESCRIPTIONNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Prepared remarks describing the FTC companion-chatbot 6(b) inquiryAUTH-33-FTC-2026-COMPANION-REMARKSCURRENT_OFFICIAL_STATEMENT_REFERENCING_OPEN_INQUIRYNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
FCC report to Congress on robocall enforcementAUTH-34-FCC-2025-ROBOCALL-CONGRESS-REPORTPRIMARY_AGENCY_REPORT_TO_CONGRESSNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
New Robodebt class action settlementAUTH-35-ROBODEBT-2026-SETTLEMENTOFFICIAL_NOTICE_OF_FEDERAL_COURT_APPROVED_SETTLEMENTNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Income Apportionment Resolution SchemeAUTH-36-ROBODEBT-INCOME-APPORTIONMENT-RESOLUTIONPRIMARY_OPERATOR_REMEDY_PROGRAM_DESCRIPTIONNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Operation Myrtleford investigation reportAUTH-37-ROBODEBT-NACC-MYRTLEFORDPRIMARY_INDEPENDENT_INVESTIGATION_REPORTINDEPENDENT_FOR_LIMITED_PROPOSITIONPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Strengthening Multimedia Integrity in the Generative AI EraAUTH-38-C2PA-NSA-GUIDANCEMULTI_AGENCY_TECHNICAL_GUIDANCENOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Verifying Provenance of Digital Media: Why the C2PA Security Model Is Not Yet Ready for High-Stakes UseAUTH-39-C2PA-INDEPENDENT-FORMAL-ANALYSISINDEPENDENT_FORMAL_METHODS_RESEARCHINDEPENDENT_FOR_LIMITED_PROPOSITIONPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingPREPRINT_CURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
New framework for auditing machine unlearningAUTH-40-GOOGLE-UNLEARNING-AUDIT-2026PRIMARY_RESEARCH_SUMMARYPARTIAL_OR_UNRESOLVEDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Unlearning or Obfuscating? Jogging the Memory of Unlearned LLMs via Benign RelearningAUTH-41-ICLR-BENIGN-RELEARNINGPEER_REVIEWED_EMPIRICAL_SECURITY_EVALUATIONINDEPENDENT_FOR_LIMITED_PROPOSITIONHIGH_FOR_STATED_PROCEDUREPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Automated Emotion Recognition in the WorkplaceAUTH-42-ACM-WORKPLACE-EMOTION-RECOGNITIONPEER_REVIEWED_ANALYSIS_OF_PROPOSED_SYSTEMSINDEPENDENT_FOR_LIMITED_PROPOSITIONPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Emotion AI in Job Interviews: Injustice, Emotional Labor, and Design ImplicationsAUTH-43-ACM-JOB-INTERVIEW-EMOTION-AIPEER_REVIEWED_AFFECTED_USER_STUDYINDEPENDENT_FOR_LIMITED_PROPOSITIONPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Russian War Report: hacked news program and Zelenskyy surrender deepfakeAUTH-44-UKRAINE-DFRLAB-ZELENSKYYINDEPENDENT_OSINT_INCIDENT_DOCUMENTATIONINDEPENDENT_FOR_LIMITED_PROPOSITIONPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Do deepfake videos undermine our epistemic trust?AUTH-45-UKRAINE-PLOS-DEEPFAKE-TRUSTPEER_REVIEWED_THEMATIC_ANALYSISINDEPENDENT_FOR_LIMITED_PROPOSITIONPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
VIGINUM 2024 activity reportAUTH-46-VIGINUM-ACTIVITY-2024PRIMARY_OPERATOR_TRANSPARENCY_REPORTNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Opinion of the VIGINUM Ethics and Scientific Committee on 2024 activityAUTH-47-VIGINUM-ETHICS-2024-OPINIONPUBLIC_OVERSIGHT_COMMITTEE_OPINIONINDEPENDENT_FOR_LIMITED_PROPOSITIONPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Protection of public debate during the March 2026 municipal electionsAUTH-48-VIGINUM-MUNICIPAL-2026PRIMARY_OPERATOR_ELECTION_REPORTNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Federal Trade Commission oversight testimony describing the companion-chatbot 6(b) studyAUTH-49-FTC-2026-OVERSIGHT-TESTIMONYPRIMARY_AGENCY_CURRENTNESS_RECORDNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENTNESS_SUPPLEMENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
FCC Agency Financial Report recording referral of the Steve Kramer forfeiture to DOJ for collectionAUTH-50-FCC-FY2024-FINANCIAL-REPORTPRIMARY_AGENCY_POST_ORDER_STATUS_RECORDNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingPOST_ORDER_REFERRAL_RECORDED_COLLECTION_NOT_RETRIEVEDNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
C2PA 2.4 explainer and durable Content Credentials guidanceAUTH-51-C2PA-2-4-EXPLAINERPRIMARY_STANDARD_SCOPE_AND_LIMITATION_RECORDPARTIAL_OR_UNRESOLVEDHIGH_FOR_STATED_PROCEDUREPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_SPECIFICATION_VERSION_ACCESSED_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Auditing Machine Unlearning: A Systematic Research on Whether Existing Algorithms Truly ForgetAUTH-52-UNLEARNING-SYSTEMATIC-AUDIT-2026INDEPENDENT_METHODS_AND_LIMITATIONS_STUDYINDEPENDENT_FOR_LIMITED_PROPOSITIONHIGH_FOR_STATED_PROCEDUREPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_RESEARCH_AS_OF_2026-07-28_NOT_PEER_REVIEW_CERTIFICATIONNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
VIGINUM mission page recording the 2026 decree that strengthened the service’s remitAUTH-53-VIGINUM-2026-MISSION-DECREEPRIMARY_OPERATOR_AND_LEGAL_CURRENTNESS_RECORDNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingCURRENT_AS_OF_2026-07-28NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Allegheny County FY 2026–27 Needs-Based Plan and Budget describing continued AFST useAUTH-54-AFST-FY2026-27-BUDGETPRIMARY_OPERATOR_CURRENTNESS_RECORDNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingACTIVE_OPERATOR_DESCRIPTION_FY2026_27NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
FluSight final forecast release for the 2025–2026 influenza seasonAUTH-55-FLUSIGHT-2025-26-FINAL-RELEASEPRIMARY_OPERATOR_SEASON_CURRENTNESS_RECORDNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizing2025_26_SEASON_COMPLETE_PROGRAM_CONTINUESNOT_AUTOMATICALLY_PRESENT; source must explicitly include it
Home Office response confirming continued use and published guidance for CARSAUTH-56-CARS-2026-HOME-OFFICE-RESPONSEPRIMARY_OPERATOR_CURRENTNESS_AND_SCOPE_RECORDNOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORDPARTIAL_PUBLIC_METHODSPUBLIC_RECORD_OR_SUMMARY_ONLYNOT_APPLICABLE_OR_SOURCE_SPECIFIC; review the linked record before generalizingACTIVE_AS_OF_2026-06-25NOT_AUTOMATICALLY_PRESENT; source must explicitly include it
56Read source-specific blind spots and correction mechanismsNO OPAQUE COMPOSITE SCORE

AUTH-01-FTC-COMPANION-INQUIRY · FTC inquiry into AI chatbots acting as companions

Scope: United States — federal — The FTC issued compulsory Section 6(b) orders concerning companion-chatbot safety, children and teens, disclosures, monetization, and data handling.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: A completed enforcement finding; Clinical causation; Product-specific liability or prevalence of harm

Correction mechanism: FTC modifies, closes, or publishes findings from the inquiry.

AUTH-02-FCC-BIDEN-ROBOCALL-FINE · FCC final action concerning AI-generated Biden robocalls

Scope: United States — federal — The FCC imposed a $6 million fine on Steve Kramer concerning thousands of spoofed robocalls using an AI-generated voice resembling President Biden and discouraging primary voting.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: How many recipients believed the call; Any measurable change in voting behavior; Broader strategic election effect

Correction mechanism: FCC or a reviewing court changes the disposition.

AUTH-03-META-OVERSIGHT-AYAHUASCA · Meta Oversight Board decision: Ayahuasca brew

Scope: Platform governance — global service context — The Board reviewed a removal involving ayahuasca-related content, found rule-notice and necessity problems, overturned the removal, and required restoration.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: Restoration of prior reach or recommendation position; Correction of all account or ranking signals; Compensation or complete downstream repair

Correction mechanism: The Board or Meta revises the public decision record.

AUTH-04-ROBODEBT-ROYAL-COMMISSION · Report of the Royal Commission into the Robodebt Scheme

Scope: Australia — federal — The Royal Commission documented the design, administration, legality, governance failures, and human consequences of the Robodebt scheme.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: That every individual harm or death had one exclusive cause; That all automated public-benefit systems share the same design; A universal remedy outcome

Correction mechanism: The Commission record or government response is formally corrected.

AUTH-05-EU-AI-ACT · Regulation (EU) 2024/1689 — Artificial Intelligence Act

Scope: European Union — The AI Act establishes risk-based obligations and specified prohibitions, including bounded restrictions on certain emotion-inference uses in workplaces and educational institutions.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: A universal cognitive-liberty code; Uniform implementation outside the EU; A scientific finding that all emotion inference is invalid

Correction mechanism: A corrigendum, amendment, delegated act, implementing act, or authoritative judgment changes interpretation.

AUTH-06-COLORADO-NEURAL-DATA · Colorado HB24-1058 — Protect Privacy of Biological Data

Scope: Colorado, United States — Colorado expanded consumer privacy protections to specified biological and neural data within the state-law framework.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: A federal U.S. neurorights regime; Universal mental-privacy protection; How every neural-data product will be enforced

Correction mechanism: Colorado amends, repeals, or materially reinterprets the law.

AUTH-07-C2PA-SPEC-2-4 · C2PA Technical Specification 2.4

Scope: International technical standard — C2PA defines signed manifests, claims, assertions, ingredients, and trust mechanisms for content provenance.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: Truth of the depicted event; That unsigned media is fake; That credentials survive every transformation or platform

Correction mechanism: C2PA publishes a superseding specification or corrigendum.

AUTH-08-NIST-MACHINE-UNLEARNING · NIST glossary entry: Machine Unlearning

Scope: United States — technical reference — NIST defines machine unlearning as removing the influence of selected training data from a trained machine-learning model.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: That every unlearning method is exact; That output blocking removes model influence; Legal compliance for any specific implementation

Correction mechanism: NIST revises or supersedes the definition.

AUTH-09-SISA-UNLEARNING · Machine Unlearning (SISA training architecture)

Scope: Research literature — The SISA approach limits the influence of individual data points to support more efficient retraining-based unlearning.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: Perfect unlearning for every model class; Legal sufficiency; Removal from downstream vendors, logs, or backups

Correction mechanism: The authors publish a correction or retraction.

AUTH-10-BARRETT-FACIAL-MOVEMENTS · Emotional Expressions Reconsidered: Challenges to Inferring Emotion From Human Facial Movements

Scope: Scientific literature — The review finds that facial movements are not universal, diagnostic fingerprints from which emotion can be reliably inferred across contexts.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: That facial movement detection itself is impossible; That every affective-computing application has zero utility; Validity of other modalities without separate evidence

Correction mechanism: The journal issues a correction or retraction.

AUTH-11-CHICAGO-PREDICTIVE-RISK · Advisory Concerning the Chicago Police Department’s Predictive Risk Models

Scope: Chicago, Illinois, United States — The OIG documented CPD person-level predictive risk models, including the Strategic Subjects List and Crime and Victimization Risk Model, and recorded their decommissioning on November 1, 2019.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: That every predictive-policing system uses the same model; A single causal effect on crime or public trust; Complete downstream deletion of historical scores

Correction mechanism: Chicago OIG corrects the advisory.

AUTH-12-ALLEGHENY-FAMILY-SCREENING · Allegheny Family Screening Tool

Scope: Allegheny County, Pennsylvania, United States — The county describes a screening score estimating the chance of future out-of-home placement; high scores can require screening in, while other scores supplement human judgment.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Independent accuracy, fairness, or causal benefit; That human judgment neutralizes automation bias; Complete remedy for errors

Correction mechanism: The county changes the model, threshold, mandate, or public description.

AUTH-13-SYRI-JUDGMENT · District Court of The Hague judgment concerning SyRI

Scope: Netherlands — The court held that the SyRI legal framework, as applied and balanced in the case, was incompatible with Article 8 of the European Convention on Human Rights.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: That all fraud-risk analytics are unlawful; A universal rule outside the judgment’s jurisdiction and facts; Complete remediation for affected communities

Correction mechanism: A higher court, official correction, or superseding law changes the status.

AUTH-14-CDC-FLUSIGHT · CDC FluSight forecasting

Scope: United States — public health — CDC uses ensemble influenza forecasts to support planning and publishes evaluation information after seasons.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Perfect forecast accuracy; Authority to take coercive action against individuals; Transferability to political or security prediction

Correction mechanism: CDC changes the forecast methodology or program status.

AUTH-15-HOME-OFFICE-CARS · Home Office Complexity Application Routing Solution — Visits

Scope: United Kingdom — The Home Office describes a rules-based system that routes visit visa applications by assessed complexity and states that it does not decide applications.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Absence of automation bias; Fairness or accuracy across groups; No effect on processing time or applicant burden

Correction mechanism: The Home Office updates or withdraws the record.

AUTH-16-UNHCR-DISPLACEMENT-FORECAST · UNHCR predictive analytics for forced displacement

Scope: International humanitarian planning — UNHCR documents predictive analytics and forecasting work intended to inform humanitarian planning for forced displacement.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Reliable prediction of individual movement; Automatic entitlement or enforcement decisions; Universal model performance across crises

Correction mechanism: UNHCR corrects or supersedes the technical record.

AUTH-17-EU-JRC-DYNAMIC-CONFLICT-RISK · Dynamic Conflict Risk Model technical report

Scope: European Union — conflict early warning — The JRC presents quantitative conflict-risk modeling as an input integrated with qualitative early-warning analysis.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Certainty about conflict onset; A lawful basis for person-level intervention; Strategic effect of any action taken from a forecast

Correction mechanism: JRC issues a correction or superseding model report.

AUTH-18-META-CIB-2022 · Meta 2022 coordinated inauthentic behavior enforcement report

Scope: Platform threat reporting — Meta reported that more than two-thirds of disrupted networks likely used GAN-generated profile pictures while emphasizing behavior rather than content as the enforcement basis.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: State sponsorship from an image alone; AI authorship of all content; Audience persuasion or strategic effect

Correction mechanism: Meta corrects the report.

AUTH-19-OPENAI-COVERT-IO · Disrupting deceptive uses of AI by covert influence operations

Scope: Platform threat reporting — OpenAI reported disrupting five covert influence operations and stated it did not observe meaningful increases in audience engagement or reach attributable to its services.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: No exposure or effect anywhere; No off-platform activity; Independent attribution beyond provider evidence

Correction mechanism: OpenAI corrects or updates the report.

AUTH-20-LIARS-DIVIDEND-APSR · The Liar’s Dividend: Can Politicians Claim Misinformation to Evade Accountability?

Scope: Research literature — Across five survey experiments with more than 15,000 adults, false claims that adverse information was misinformation sometimes increased support, while deepfake claims were largely ineffective against video evidence in the tested settings.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: Effect in every country or crisis; Long-term behavior change; Effectiveness against every media format

Correction mechanism: The journal issues a correction or retraction.

AUTH-21-VIGINUM-AI-THREAT · Information threat linked to artificial intelligence

Scope: France — VIGINUM describes an official mandate focused on foreign digital interference, publicly accessible content, and manipulative behavior patterns.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: That every classification is correct; Authority over lawful domestic belief; Strategic effect of every detected campaign

Correction mechanism: VIGINUM changes its mandate or corrects the report.

AUTH-22-UKRAINE-CPD-DEEPFAKES · Analytical report on AI-generated videos used to discredit Ukraine’s Defence Forces

Scope: Ukraine — The report documents the 2022 Zelenskyy surrender deepfake and later Zaluzhnyi deepfakes, including official detection and debunking.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Precise operator identity for every artifact; Continuing audience belief; Measured operational or strategic effect

Correction mechanism: The issuing body corrects attribution or chronology.

AUTH-23-CHICAGO-NIJ-EVALUATION · Predictions Put Into Practice: A Quasi-Experimental Evaluation of Chicago’s Predictive Policing Pilot

Scope: United States — federal research publication — A quasi-experimental evaluation of the 2013 Strategic Subjects List pilot involving 426 people and measures related to gun violence.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Complete individual-level notice or remedy; That every later CPD risk model had the same implementation; Causal proof concerning all predictive-policing systems

Correction mechanism: NIJ or the underlying authors publish an erratum or corrected outcome statement.

AUTH-24-SYRI-DATA-DESTRUCTION · Parliamentary answers concerning SyRI data destruction and decision use

Scope: Netherlands — Official answers state that responsible administrative bodies made no decisions based on SyRI-derived data and that all data processed in connection with SyRI had been destroyed.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Independent forensic verification of destruction; Deletion from every backup, derivative, vendor, or institutional system; Complete repair for communities subjected to SyRI projects

Correction mechanism: The Netherlands government or a court corrects the destruction or decision-use statement.

AUTH-25-AFST-PHASE2-EVALUATION · Impact Evaluation of the Allegheny Family Screening Tool — Phase 2

Scope: Allegheny County, Pennsylvania, United States — Phase 2 evaluation of AFST implementation and surrounding policy and practice changes, including screening, service acceptance, later referrals, removals, and racial disparity analyses.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: A randomized causal estimate of the tool alone; Absence of confounding from COVID-19 or contemporaneous policy and practice changes; Complete family-level notice, appeal, correction, or downstream repair

Correction mechanism: Allegheny County or the authors publish a revised evaluation.

AUTH-26-AFST-HUMAN-RACIAL-DISPARITY · Algorithms, Humans and Racial Disparities in Child Protection Systems

Scope: Allegheny County, Pennsylvania, United States — Research examining how human screeners used AFST scores and how human decision-making changed racial disparities relative to algorithm-only recommendations.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: That all human review reduces disparities; Complete absence of disparate impact; Generalization beyond the AFST setting and period

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-27-CARS-ICIBI-INSPECTION · An inspection of visit visa operations, December 2022 to January 2023

Scope: United Kingdom — Inspection of the Complexity Application Routing Solution’s efficiency, effectiveness, consistency, equality controls, workflow effects, and assurance practices.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: Applicant-level causal effects of routing; Customer or affected-person experience; the inspection states that perspective was not examined; Complete subgroup error rates, appeals, or downstream remedy

Correction mechanism: ICIBI publishes an erratum or follow-up that changes the inspection findings.

AUTH-28-CARS-HOME-OFFICE-RESPONSE · Home Office response to inspection of visit visa operations

Scope: United Kingdom — The Home Office accepted all five recommendations and stated that CARS would be implemented permanently.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: Independent proof that every recommendation produced the intended outcome; Applicant-level fairness or remedy; Current 2026 performance without a new inspection

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-29-FLUSIGHT-2024-25-EVALUATION · FluSight 2024–2025 Season Evaluation

Scope: United States — federal public health — Season evaluation of 46 submitted models, 35 included models, ensemble performance, uncertainty intervals, and delayed publication after missed rapid growth.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: A guarantee of future forecast accuracy; A direct causal estimate of planning or health outcomes produced by the forecasts; Individual-level medical decisions

Correction mechanism: CDC publishes an erratum or corrected season evaluation.

AUTH-30-FLUSIGHT-2023-24-EVALUATION · FluSight 2023–2024 Season Evaluation

Scope: United States — federal public health — Season evaluation of ensemble and component models, including performance during rapid increases and declines.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Direct evidence of downstream resource-allocation benefit; Performance outside evaluated seasons; Individual clinical validity

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-31-UNHCR-JETSON-TECHNICAL · Project Jetson technical specifications

Scope: Somalia / UNHCR humanitarian operations — Technical description of historical displacement data, modeling inputs, and Somalia regional forecasting.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Independent out-of-sample operational impact; Affected-community benefit or harm; Current production deployment across UNHCR

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-32-UNHCR-AI-APPROACH-2025 · UNHCR AI Approach

Scope: Global / UNHCR — UNHCR policy describes a long-standing use of predictive analytics, including Jetson and other planning and nowcasting activities.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Independent model accuracy or benefit; Individual or community consent and remedy outcomes; A single unified production model

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-33-FTC-2026-COMPANION-REMARKS · Prepared remarks describing the FTC companion-chatbot 6(b) inquiry

Scope: United States — federal — March 2026 FTC remarks state that Section 6(b) orders had been issued to seven companies and describe the inquiry’s information-gathering scope.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Public findings, final enforcement, settlement, or closure; Product-specific wrongdoing; Clinical causality

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-34-FCC-2025-ROBOCALL-CONGRESS-REPORT · FCC report to Congress on robocall enforcement

Scope: United States — federal — The report records the September 30, 2024 $6 million forfeiture order against Steve Kramer and reports 2024 collection information under the relevant statutory section.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Payment, collection, modification, or judicial disposition after the period covered; Voter behavior or election outcome effects; State-level proceedings not described in the report

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-35-ROBODEBT-2026-SETTLEMENT · New Robodebt class action settlement

Scope: Australia — federal — The Federal Court approved a new settlement on June 23, 2026 and the Commonwealth committed an additional $475 million for eligible group members.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Individual complete repair; Correction of every downstream record or inference; A merits judgment on every disputed issue

Correction mechanism: The Federal Court or settlement administrator changes the approval or payment terms.

AUTH-36-ROBODEBT-INCOME-APPORTIONMENT-RESOLUTION · Income Apportionment Resolution Scheme

Scope: Australia — federal — Applications opened January 30, 2026 and close January 29, 2027; eligible people may receive a resolution payment concerning income-apportionment debts.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Take-up, accessibility, approval, or rejection rates; Complete compensation; Downstream data correction or repeat-error prevention

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-37-ROBODEBT-NACC-MYRTLEFORD · Operation Myrtleford investigation report

Scope: Australia — federal — Investigation of six referrals from the Robodebt Royal Commission; the NACC found serious corrupt conduct by two people and no corrupt conduct by four.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: Criminal conviction; Complete institutional reform; Complete repair to affected people

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-38-C2PA-NSA-GUIDANCE · Strengthening Multimedia Integrity in the Generative AI Era

Scope: International cybersecurity guidance — Guidance on content credentials, provenance, watermarking, fingerprinting, signing, and resilience against metadata stripping or modification.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Independent real-world survival rates across all platforms; Universal adoption; Proof that unsigned media is false

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-39-C2PA-INDEPENDENT-FORMAL-ANALYSIS · Verifying Provenance of Digital Media: Why the C2PA Security Model Is Not Yet Ready for High-Stakes Use

Scope: Global technical standard — Independent formal-methods and security analysis of C2PA core protocols and claimed security goals.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: That every C2PA implementation fails; Field prevalence of each vulnerability; A final standards-body disposition

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-40-GOOGLE-UNLEARNING-AUDIT-2026 · New framework for auditing machine unlearning

Scope: Global machine-learning research — A statistical framework for testing whether post-unlearning observation distributions differ in a way consistent with deletion.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: Exact deletion in deployed foundation models; Production deployment across vendors; Elimination of every memorized or inferential effect

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-41-ICLR-BENIGN-RELEARNING · Unlearning or Obfuscating? Jogging the Memory of Unlearned LLMs via Benign Relearning

Scope: Global machine-learning research — Empirical research showing that several optimization-based LLM unlearning methods were susceptible to benign relearning attacks that recovered supposedly forgotten knowledge.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: Failure of every unlearning method; Production-vendor deletion outcomes; Exact recovery of every target item

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-42-ACM-WORKPLACE-EMOTION-RECOGNITION · Automated Emotion Recognition in the Workplace

Scope: Global / workplace technology — Study of proposed workplace emotion-recognition technologies, their inputs, training, outputs, institutional actions, and implications.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: Independent construct validity for a specified deployed system; Clinical knowledge of employee emotion; Population-wide performance

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-43-ACM-JOB-INTERVIEW-EMOTION-AI · Emotion AI in Job Interviews: Injustice, Emotional Labor, and Design Implications

Scope: United States / job interviews — Study comparing perceived justice and emotional labor when interviews were evaluated by emotion AI versus human resources.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: Construct validity of the emotion score; Hiring outcome benefit or predictive validity; Generalization to every hiring system

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-44-UKRAINE-DFRLAB-ZELENSKYY · Russian War Report: hacked news program and Zelenskyy surrender deepfake

Scope: Ukraine / Russia information environment — Open-source documentation of the hacked news channel, deepfake circulation, rapid rebuttal, and observed online reactions.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: Representative recipient exposure or belief; Frontline disruption; Operator identity, sponsorship chain, behavior change, or strategic effect

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-45-UKRAINE-PLOS-DEEPFAKE-TRUST · Do deepfake videos undermine our epistemic trust?

Scope: Global Twitter discourse concerning the Russo-Ukrainian war — Thematic analysis of 4,869 tweets about deepfakes and the Russo-Ukrainian war, including skepticism and false claims that authentic media were deepfakes.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: Representative population belief; Causal behavior change; Frontline operational effect

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-46-VIGINUM-ACTIVITY-2024 · VIGINUM 2024 activity report

Scope: France — Annual report on mandate, public-data collection, four-month deletion limit, exclusions of facial and voice identification, staffing, operations, and ethics committee oversight.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Independent false-positive rate; Incident-level appeals or correction outcomes; Affected-community experience

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-47-VIGINUM-ETHICS-2024-OPINION · Opinion of the VIGINUM Ethics and Scientific Committee on 2024 activity

Scope: France — Public multidisciplinary committee opinion reviewing VIGINUM’s institutional maturation, legal framework, collection practices, and transparency obligations.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: A judicially independent appeal forum; Measured false-positive burden; Correction outcomes for specific people or organizations

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-48-VIGINUM-MUNICIPAL-2026 · Protection of public debate during the March 2026 municipal elections

Scope: France — Public report documenting four foreign digital interference operations detected and characterized during the 2026 municipal elections and the coordination network’s public reporting.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Independent verification of every attribution; False-positive, appeal, or correction outcomes; Voter belief or behavior change

Correction mechanism: The issuing body publishes a corrected, superseding, or withdrawn record.

AUTH-49-FTC-2026-OVERSIGHT-TESTIMONY · Federal Trade Commission oversight testimony describing the companion-chatbot 6(b) study

Scope: United States — federal — The Commission described the September 2025 companion-chatbot orders as an ongoing Section 6(b) study intended to understand safety practices and effects on children and teens.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: A public final report, closure, settlement, or enforcement disposition; Product-specific wrongdoing or clinical causation; The contents of confidential company responses

Correction mechanism: The FTC publishes findings, closure, enforcement, or corrected testimony.

AUTH-50-FCC-FY2024-FINANCIAL-REPORT · FCC Agency Financial Report recording referral of the Steve Kramer forfeiture to DOJ for collection

Scope: United States — federal — The FCC recorded the $6 million forfeiture order and stated that the matter had been referred to the Department of Justice for collection.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Payment, collection, settlement, or a judicial judgment collecting the forfeiture; Recipient belief or voting behavior; A final state-law disposition for every related proceeding

Correction mechanism: FCC, DOJ, or a court publishes a later collection, settlement, modification, or review record.

AUTH-51-C2PA-2-4-EXPLAINER · C2PA 2.4 explainer and durable Content Credentials guidance

Scope: Open technical standard — cross-jurisdictional — The explainer states that provenance metadata can be removed and describes durable credentials using hard and soft bindings, including watermarking and fingerprinting, to help recover a manifest.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: That credentials survive every screenshot, repost, print-scan, platform, or crisis chain; That a signed asset depicts a true event; That unsigned media is false or synthetic

Correction mechanism: C2PA revises the explanation, security model, or durability claims.

AUTH-52-UNLEARNING-SYSTEMATIC-AUDIT-2026 · Auditing Machine Unlearning: A Systematic Research on Whether Existing Algorithms Truly Forget

Scope: Research — cross-jurisdictional — The study evaluates whether existing unlearning algorithms truly erase designated influence and treats reliable auditing and residual leakage as open challenges.

Conflict-of-interest boundary: NO_CONFLICT_FINDING_MADE; relationship disclosed

Known blind spots: A universal failure of every unlearning method; Exact behavior of any named production foundation model; Legal compliance in a specific jurisdiction

Correction mechanism: The paper is withdrawn, corrected, or superseded by a stronger production audit.

AUTH-53-VIGINUM-2026-MISSION-DECREE · VIGINUM mission page recording the 2026 decree that strengthened the service’s remit

Scope: France — The official page describes VIGINUM’s principal mission as defensive and states that Decree No. 2026-70 of 11 February 2026 strengthened its missions.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Independent validation of attribution decisions; A public false-positive rate or complete correction and appeal system; Affected-community acceptance or behavior effects

Correction mechanism: The decree, mission page, or institutional remit is amended or superseded.

AUTH-54-AFST-FY2026-27-BUDGET · Allegheny County FY 2026–27 Needs-Based Plan and Budget describing continued AFST use

Scope: Allegheny County, Pennsylvania, United States — The county plan continues to describe the AFST as a data-driven model used in child-welfare call-screening decisions.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Independent current calibration or subgroup validity; Complete notice, appeal, correction, or downstream repair outcomes; Causal improvement in child or family outcomes

Correction mechanism: The county retires, replaces, materially changes, or corrects the tool’s current status.

AUTH-55-FLUSIGHT-2025-26-FINAL-RELEASE · FluSight final forecast release for the 2025–2026 influenza season

Scope: United States — federal public health — The CDC marked the June 11 release as the final influenza forecast release for the 2025–2026 season and stated that reporting would resume in fall 2026.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Uniform accuracy at every horizon or during rapid trend changes; A person-level risk decision system; Causal effect on any particular public-health intervention

Correction mechanism: CDC changes the season status, evaluation, or program architecture.

AUTH-56-CARS-2026-HOME-OFFICE-RESPONSE · Home Office response confirming continued use and published guidance for CARS

Scope: United Kingdom — The Home Office described CARS as a routing tool for identifying application complexity and confirmed that visit and student guidance documents were public.

Conflict-of-interest boundary: SUBJECT_INSTITUTION_OR_OPERATOR_INTEREST_PRESENT

Known blind spots: Independent subgroup error rates or fairness; Absence of automation bias in downstream scrutiny; Complete applicant notice, appeal, or remedy outcomes

Correction mechanism: The Home Office retires, replaces, or materially changes CARS.

45 VERSIONED SEARCH RECORDS

Five generations of nine bounded searches, including the WIP.62 version-5 refresh

Every refresh points to its prior search, records whether it confirms, supplements, narrows, or supersedes the earlier result, and states whether public wording changed.

  1. v1FTC companion-chatbot inquiry outcomeHISTORICAL_WIP57_RECORD
    Search ID
    SEARCH-01-FTC-COMPANION
    Prior search
    NONE
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: OPEN_INQUIRY_NO_PUBLIC_FINDINGS_RETRIEVED_AS_OF_2026-07-28

    Known blind spots: Nonpublic company responses and confidential FTC work are inaccessible.

    Dependent claim: Public wording that the FTC is investigating companion safety.

    Prohibited inference: Do not describe an inquiry as a finding or enforcement action.

    Queries and repositories

    Queries

    • site:ftc.gov 2026 AI chatbots acting as companions inquiry findings
    • site:ftc.gov companion products Section 6(b) report enforcement settlement

    Repositories

    • Federal Trade Commission press releases, reports, cases, speeches, and testimony

    Correction trigger: FTC publishes a public report or corrects the status.

    Reopening trigger: Search again when a 6(b) report, case, settlement, closure notice, or testimony appears.

  2. v1FCC Biden robocall payment, review, and follow-on dispositionHISTORICAL_WIP57_RECORD
    Search ID
    SEARCH-02-FCC-ROBOCALL
    Prior search
    NONE
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: FINAL_ORDER_RETRIEVED_COLLECTION_OR_LATER_JUDICIAL_DISPOSITION_NOT_RETRIEVED

    Known blind spots: Collection can proceed through channels not indexed in the bounded search.

    Dependent claim: The incident produced a final federal administrative penalty.

    Prohibited inference: Do not call the penalty paid, collected, reversed, or judicially affirmed without a source.

    Queries and repositories

    Queries

    • site:docs.fcc.gov Steve Kramer $6,000,000 forfeiture paid collected judicial review
    • site:justice.gov Steve Kramer FCC robocall forfeiture collection

    Repositories

    • FCC orders and reports to Congress; DOJ and federal court records

    Correction trigger: FCC, DOJ, or a court publishes a disposition.

    Reopening trigger: Repeat FCC, DOJ, PACER/public court, and state-attorney-general searches.

  3. v1Meta ayahuasca downstream restorationHISTORICAL_WIP57_RECORD
    Search ID
    SEARCH-03-META-AYAHUASCA
    Prior search
    NONE
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: POST_RESTORATION_STRIKE_RANKING_REACH_REPAIR_NOT_RETRIEVED

    Known blind spots: Internal ranking and account-state systems are not publicly observable.

    Dependent claim: A restored post is one remedy stage.

    Prohibited inference: Do not equate visible restoration with complete repair.

    Queries and repositories

    Queries

    • site:oversightboard.com ayahuasca brew implementation restoration strike ranking recommendation
    • site:transparency.meta.com ayahuasca oversight board implementation

    Repositories

    • Meta Oversight Board decisions, implementation updates, and transparency materials

    Correction trigger: Meta or the Oversight Board publishes system-level repair evidence.

    Reopening trigger: Seek implementation reports, account-state records, creator evidence, and independent audits.

  4. v1Robodebt compensation, accountability, data repair, and repeat preventionHISTORICAL_WIP57_RECORD
    Search ID
    SEARCH-04-ROBODEBT
    Prior search
    NONE
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: PARTIAL_REPAIR_AND_ACCOUNTABILITY_DOCUMENTED_COMPLETE_REPAIR_NOT_ESTABLISHED

    Known blind spots: Ongoing scheme administration and individual circumstances are not yet fully reported.

    Dependent claim: Robodebt produced large-scale harm and multiple remedy mechanisms.

    Prohibited inference: Do not label settlement or payment as complete repair.

    Queries and repositories

    Queries

    • site:servicesaustralia.gov.au 2026 Robodebt settlement resolution scheme
    • site:nacc.gov.au Robodebt investigation report 2026
    • site:ag.gov.au automated decision making Robodebt recommendations

    Repositories

    • Services Australia, Federal Court settlement notices, NACC, Attorney-General’s Department, Royal Commission

    Correction trigger: Court, administrator, agency, or audit publishes outcome and downstream-correction evidence.

    Reopening trigger: Track payment completion, exclusions, complaints, automated-decision reforms, data correction, and repeat-error metrics.

  5. v1C2PA adoption and credential survival in crisis chainsHISTORICAL_WIP57_RECORD
    Search ID
    SEARCH-05-C2PA
    Prior search
    NONE
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: LAYERED_GUIDANCE_AND_SECURITY_ANALYSIS_RETRIEVED_REAL_CRISIS_CHAIN_SURVIVAL_NOT_RETRIEVED

    Known blind spots: Platform implementations, re-encoding, screenshotting, and visible UI adoption vary and are incompletely public.

    Dependent claim: Content credentials are one authentication layer.

    Prohibited inference: Do not treat credential presence as truth or absence as falsity.

    Queries and repositories

    Queries

    • C2PA independent evaluation metadata stripping crisis distribution chain
    • site:media.defense.gov content credentials stripping fingerprint watermark C2PA
    • formal analysis C2PA security model 2026

    Repositories

    • C2PA specification, NSA/CISA partner guidance, independent formal-methods research

    Correction trigger: A peer-reviewed cross-platform field evaluation or corrected specification becomes available.

    Reopening trigger: Test signed assets across capture, editing, messaging, social platforms, screenshots, downloads, and archive workflows.

  6. v1Foundation-model machine-unlearning exactnessHISTORICAL_WIP57_RECORD
    Search ID
    SEARCH-06-UNLEARNING
    Prior search
    NONE
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: AUDIT_AND_REVERSIBILITY_RESEARCH_RETRIEVED_DEPLOYED_EXACTNESS_NOT_RETRIEVED

    Known blind spots: Production models and training data are proprietary; exact retraining references are rarely available.

    Dependent claim: Claims about machine unlearning require method- and audit-specific evidence.

    Prohibited inference: Do not describe UI deletion, refusal, or model editing as exact unlearning.

    Queries and repositories

    Queries

    • machine unlearning deployed foundation model external audit exactness 2026
    • ICLR benign relearning unlearned LLM
    • Google Research auditing machine unlearning 2026

    Repositories

    • Google Research, ICLR proceedings, NIST, peer-reviewed and primary research

    Correction trigger: A production audit with retraining reference or bounded approximation guarantee is published.

    Reopening trigger: Search benchmark, vendor transparency, regulatory audit, and independent replication records.

  7. v1Mental-state inference independent field validityHISTORICAL_WIP57_RECORD
    Search ID
    SEARCH-07-MENTAL-STATE
    Prior search
    NONE
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: DEPLOYMENT_AND_AFFECTED_USER_RESEARCH_RETRIEVED_CONSTRUCT_SPECIFIC_FIELD_VALIDITY_NOT_RETRIEVED

    Known blind spots: Proprietary systems, changing products, and weak ground truth limit evaluation.

    Dependent claim: Observed signals are not the mind itself.

    Prohibited inference: Do not convert facial, vocal, gaze, neural, or interaction features into psychological fact without construct-valid evidence.

    Queries and repositories

    Queries

    • independent field validation emotion recognition workplace construct validity consequential decisions
    • ACM emotion AI job interviews justice field study
    • emotion recognition calibration base rates subgroup performance deployment

    Repositories

    • Peer-reviewed psychology and HCI, EU law, regulator and vendor evaluation records

    Correction trigger: A preregistered, independent, consequential field validation becomes available.

    Reopening trigger: Seek system-specific model cards, threshold metrics, subgroup performance, actions, appeals, and outcomes.

  8. v1Ukraine deepfake recipient-level and strategic effectsHISTORICAL_WIP57_RECORD
    Search ID
    SEARCH-08-UKRAINE-DEEPFAKE
    Prior search
    NONE
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: ARTIFACT_DISTRIBUTION_AND_DISCOURSE_RETRIEVED_RECIPIENT_BEHAVIOR_AND_STRATEGIC_EFFECT_NOT_RETRIEVED

    Known blind spots: Military communications and platform exposure data may be nonpublic or unavailable.

    Dependent claim: The 2022 surrender deepfake was rapidly debunked and poorly received in observed online contexts.

    Prohibited inference: Do not convert ridicule or removal into proof of no recipient effect.

    Queries and repositories

    Queries

    • Zelensky surrender deepfake recipient survey belief frontline disruption effect
    • Ukraine deepfake military communications operational impact study
    • Zelensky deepfake operator attribution evidence

    Repositories

    • DFRLab, Ukrainian government records, peer-reviewed research, platform and military records

    Correction trigger: Recipient surveys, military records, platform data, or attribution evidence becomes available.

    Reopening trigger: Search Ukrainian-language records, military archives, platform transparency, surveys, and oral-history evidence.

  9. v1VIGINUM false positives, oversight outcomes, and affected-community experienceHISTORICAL_WIP57_RECORD
    Search ID
    SEARCH-09-VIGINUM
    Prior search
    NONE
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: MANDATE_OVERSIGHT_AND_ACTIVITY_RETRIEVED_FALSE_POSITIVE_AND_REMEDY_OUTCOMES_NOT_RETRIEVED

    Known blind spots: Some intelligence methods and corrections may be nonpublic.

    Dependent claim: VIGINUM provides one behavior- and foreign-interference-focused democratic-defense model with published oversight features.

    Prohibited inference: Do not treat an institutional attribution as self-proving or infer zero false positives from absent public corrections.

    Queries and repositories

    Queries

    • site:sgdsn.gouv.fr VIGINUM false positive correction appeal affected community audit
    • site:sgdsn.gouv.fr VIGINUM comité éthique scientifique avis rapport activité
    • site:legifrance.gouv.fr VIGINUM décret 2026 oversight

    Repositories

    • SGDSN/VIGINUM, ethics and scientific committee, Legifrance, election reports

    Correction trigger: An independent audit, court case, complaint record, correction notice, or affected-community study becomes available.

    Reopening trigger: Repeat French-language searches across SGDSN, Legifrance, CNIL, courts, parliament, press, and civil society.

  10. v2FTC companion-chatbot inquiry outcomeSUPPLEMENTS
    Search ID
    SEARCH-10-FTC-COMPANION-V2
    Prior search
    SEARCH-01-FTC-COMPANION
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: ONGOING_STUDY_RECONFIRMED_NO_PUBLIC_FINAL_FINDINGS_RETRIEVED_AS_OF_2026-07-28

    Known blind spots: Nonpublic submissions, internal logs, confidential deliberations, private datasets, and unindexed records remain outside the bounded search.

    Dependent claim: FTC companion-chatbot inquiry outcome

    Prohibited inference: Do not infer nonexistence, complete repair, validity, causation, actor identity, sponsorship, behavior change, or strategic effect beyond the retrieved records.

    Queries and repositories

    Queries

    • site:ftc.gov 2026 testimony companion chatbots inquiry findings
    • site:ftc.gov AI chatbots acting as companions 6(b) report enforcement settlement closure

    Repositories

    • Federal Trade Commission press releases, testimony, reports, cases, and legal library

    Correction trigger: A primary, independently attributable, or corrected record changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, field study, correction, or affected-person record appears.

  11. v2FCC Biden robocall payment, review, and follow-on dispositionSUPPLEMENTS
    Search ID
    SEARCH-11-FCC-ROBOCALL-V2
    Prior search
    SEARCH-02-FCC-ROBOCALL
    Search date
    2026-07-28
    Public wording changed
    YES

    Result: DOJ_COLLECTION_REFERRAL_RETRIEVED_PAYMENT_OR_JUDICIAL_COLLECTION_NOT_RETRIEVED

    Known blind spots: Nonpublic submissions, internal logs, confidential deliberations, private datasets, and unindexed records remain outside the bounded search.

    Dependent claim: FCC Biden robocall payment, review, and follow-on disposition

    Prohibited inference: Do not infer nonexistence, complete repair, validity, causation, actor identity, sponsorship, behavior change, or strategic effect beyond the retrieved records.

    Queries and repositories

    Queries

    • site:fcc.gov Steve Kramer robocall forfeiture DOJ collection payment
    • site:justice.gov Steve Kramer FCC forfeiture collection
    • Steve Kramer robocall forfeiture judicial review payment

    Repositories

    • FCC enforcement, agency financial reports, DOJ releases, and federal court records

    Correction trigger: A primary, independently attributable, or corrected record changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, field study, correction, or affected-person record appears.

  12. v2Meta ayahuasca downstream restorationUNCHANGED
    Search ID
    SEARCH-12-META-AYAHUASCA-V2
    Prior search
    SEARCH-03-META-AYAHUASCA
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: POST_RESTORATION_RANKING_STRIKE_RECOMMENDATION_AND_LOST_REACH_EVIDENCE_NOT_RETRIEVED

    Known blind spots: Nonpublic submissions, internal logs, confidential deliberations, private datasets, and unindexed records remain outside the bounded search.

    Dependent claim: Meta ayahuasca downstream restoration

    Prohibited inference: Do not infer nonexistence, complete repair, validity, causation, actor identity, sponsorship, behavior change, or strategic effect beyond the retrieved records.

    Queries and repositories

    Queries

    • site:oversightboard.com ayahuasca brew restoration ranking strike recommendation eligibility
    • site:transparency.meta.com ayahuasca oversight board implementation

    Repositories

    • Meta Oversight Board decisions and implementation updates; Meta transparency reporting

    Correction trigger: A primary, independently attributable, or corrected record changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, field study, correction, or affected-person record appears.

  13. v2Robodebt compensation, accountability, data repair, and repeat preventionSUPPLEMENTS
    Search ID
    SEARCH-13-ROBODEBT-V2
    Prior search
    SEARCH-04-ROBODEBT
    Search date
    2026-07-28
    Public wording changed
    YES

    Result: ADDITIONAL_COMPENSATION_AND_RESOLUTION_SCHEME_IMPLEMENTATION_RETRIEVED_COMPLETE_DOWNSTREAM_REPAIR_NOT_ESTABLISHED

    Known blind spots: Nonpublic submissions, internal logs, confidential deliberations, private datasets, and unindexed records remain outside the bounded search.

    Dependent claim: Robodebt compensation, accountability, data repair, and repeat prevention

    Prohibited inference: Do not infer nonexistence, complete repair, validity, causation, actor identity, sponsorship, behavior change, or strategic effect beyond the retrieved records.

    Queries and repositories

    Queries

    • site:servicesaustralia.gov.au robodebt settlement 2026 compensation
    • site:servicesaustralia.gov.au income apportionment resolution scheme 2026
    • site:nacc.gov.au Operation Myrtleford robodebt 2026

    Repositories

    • Services Australia, Federal Court public information, National Anti-Corruption Commission, and official implementation records

    Correction trigger: A primary, independently attributable, or corrected record changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, field study, correction, or affected-person record appears.

  14. v2C2PA adoption and credential survival in crisis chainsSUPPLEMENTS
    Search ID
    SEARCH-14-C2PA-V2
    Prior search
    SEARCH-05-C2PA
    Search date
    2026-07-28
    Public wording changed
    YES

    Result: STANDARD_CONFIRMS_METADATA_REMOVAL_AND_SOFT_BINDING_RECOVERY_REPRESENTATIVE_CRISIS_SURVIVAL_RATES_NOT_RETRIEVED

    Known blind spots: Nonpublic submissions, internal logs, confidential deliberations, private datasets, and unindexed records remain outside the bounded search.

    Dependent claim: C2PA adoption and credential survival in crisis chains

    Prohibited inference: Do not infer nonexistence, complete repair, validity, causation, actor identity, sponsorship, behavior change, or strategic effect beyond the retrieved records.

    Queries and repositories

    Queries

    • site:spec.c2pa.org 2.4 durable Content Credentials metadata removed watermark fingerprint
    • C2PA credential survival screenshot repost print scan crisis distribution study

    Repositories

    • C2PA specification, explainer, implementation guidance, independent formal and field research

    Correction trigger: A primary, independently attributable, or corrected record changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, field study, correction, or affected-person record appears.

  15. v2Foundation-model machine-unlearning exactnessSUPPLEMENTS
    Search ID
    SEARCH-15-UNLEARNING-V2
    Prior search
    SEARCH-06-UNLEARNING
    Search date
    2026-07-28
    Public wording changed
    YES

    Result: AUDIT_RESEARCH_RETRIEVED_TRUE_FORGETTING_REMAINS_OPEN_PRODUCTION_EXACTNESS_NOT_RETRIEVED

    Known blind spots: Nonpublic submissions, internal logs, confidential deliberations, private datasets, and unindexed records remain outside the bounded search.

    Dependent claim: Foundation-model machine-unlearning exactness

    Prohibited inference: Do not infer nonexistence, complete repair, validity, causation, actor identity, sponsorship, behavior change, or strategic effect beyond the retrieved records.

    Queries and repositories

    Queries

    • 2026 machine unlearning systematic audit truly forget foundation model
    • site:research.google 2026 auditing machine unlearning
    • production foundation model unlearning exact retraining equivalence audit

    Repositories

    • Google Research, arXiv and conference proceedings, standards and deployed-model documentation

    Correction trigger: A primary, independently attributable, or corrected record changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, field study, correction, or affected-person record appears.

  16. v2Mental-state inference independent field validityUNCHANGED
    Search ID
    SEARCH-16-MENTAL-STATE-V2
    Prior search
    SEARCH-07-MENTAL-STATE
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: NO_COMPLETE_INDEPENDENT_CONSTRUCT_SPECIFIC_FIELD_VALIDATION_RETRIEVED

    Known blind spots: Nonpublic submissions, internal logs, confidential deliberations, private datasets, and unindexed records remain outside the bounded search.

    Dependent claim: Mental-state inference independent field validity

    Prohibited inference: Do not infer nonexistence, complete repair, validity, causation, actor identity, sponsorship, behavior change, or strategic effect beyond the retrieved records.

    Queries and repositories

    Queries

    • independent field validation emotion recognition named construct calibration base rate subgroup 2026
    • workplace emotion AI independent field study construct validity outcome
    • border deception detection field validation base rate subgroup

    Repositories

    • Peer-reviewed HCI and affective-science literature, regulator records, deployment audits, and operator documentation

    Correction trigger: A primary, independently attributable, or corrected record changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, field study, correction, or affected-person record appears.

  17. v2Ukraine deepfake recipient-level and strategic effectsUNCHANGED
    Search ID
    SEARCH-17-UKRAINE-DEEPFAKE-V2
    Prior search
    SEARCH-08-UKRAINE-DEEPFAKE
    Search date
    2026-07-28
    Public wording changed
    YES

    Result: ARTIFACT_AND_RAPID_RESPONSE_RETRIEVED_RECIPIENT_BELIEF_FRONTLINE_BEHAVIOR_AND_STRATEGIC_EFFECT_NOT_RETRIEVED

    Known blind spots: Nonpublic submissions, internal logs, confidential deliberations, private datasets, and unindexed records remain outside the bounded search.

    Dependent claim: Ukraine deepfake recipient-level and strategic effects

    Prohibited inference: Do not infer nonexistence, complete repair, validity, causation, actor identity, sponsorship, behavior change, or strategic effect beyond the retrieved records.

    Queries and repositories

    Queries

    • Zelensky surrender deepfake recipient belief frontline behavior study
    • Ukraine Zelensky deepfake operational effect soldiers survey
    • Zelenskyy deepfake trust study recipient exposure

    Repositories

    • Incident forensics, Ukrainian official records, independent journalism, peer-reviewed deepfake and trust research

    Correction trigger: A primary, independently attributable, or corrected record changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, field study, correction, or affected-person record appears.

  18. v2VIGINUM false positives, oversight outcomes, and affected-community experienceSUPPLEMENTS
    Search ID
    SEARCH-18-VIGINUM-V2
    Prior search
    SEARCH-09-VIGINUM
    Search date
    2026-07-28
    Public wording changed
    YES

    Result: MISSION_CURRENTNESS_RETRIEVED_PUBLIC_FALSE_POSITIVE_REMEDY_AND_AFFECTED_COMMUNITY_REGISTER_NOT_RETRIEVED

    Known blind spots: Nonpublic submissions, internal logs, confidential deliberations, private datasets, and unindexed records remain outside the bounded search.

    Dependent claim: VIGINUM false positives, oversight outcomes, and affected-community experience

    Prohibited inference: Do not infer nonexistence, complete repair, validity, causation, actor identity, sponsorship, behavior change, or strategic effect beyond the retrieved records.

    Queries and repositories

    Queries

    • site:sgdsn.gouv.fr VIGINUM 2026 decree mission false positive correction appeal
    • site:sgdsn.gouv.fr VIGINUM ethics committee false positives remedy
    • VIGINUM affected community appeal correction false positive

    Repositories

    • SGDSN/VIGINUM reports and mission pages, ethics committee opinions, legal records, oversight and affected-community publications

    Correction trigger: A primary, independently attributable, or corrected record changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, field study, correction, or affected-person record appears.

  19. v3FTC companion-chatbot inquiry outcomeCONFIRMS
    Search ID
    SEARCH-19-FTC-COMPANION-V3
    Prior search
    SEARCH-10-FTC-COMPANION-V2
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: OPEN_INQUIRY_RECONFIRMED_NO_PUBLIC_FINAL_FINDINGS_OR_CLOSURE_RETRIEVED

    Known blind spots: Nonpublic company responses and confidential FTC work are inaccessible.

    Dependent claim: Public wording that the FTC inquiry exists and remained open in retrieved records.

    Prohibited inference: Do not describe an inquiry as a finding, liability determination, prevalence estimate, or enforcement outcome.

    Queries and repositories

    Queries

    • site:ftc.gov 2026 companion chatbot inquiry findings closure report
    • site:ftc.gov testimony AI companions Section 6(b) 2026

    Repositories

    • Federal Trade Commission press releases, testimony, reports, cases, and legal library

    Correction trigger: A primary, independently attributable, adjudicative, or corrected record materially changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, implementation record, independent field study, correction, or affected-person record appears.

  20. v3FCC Biden robocall payment, collection, review, and court statusSUPPLEMENTS
    Search ID
    SEARCH-20-FCC-ROBOCALL-V3
    Prior search
    SEARCH-11-FCC-ROBOCALL-V2
    Search date
    2026-07-28
    Public wording changed
    YES

    Result: FINAL_FORFEITURE_AND_DOJ_COLLECTION_REFERRAL_RETRIEVED_PAYMENT_OR_COURT_DISPOSITION_NOT_RETRIEVED

    Known blind spots: Collection or litigation can proceed through records not indexed in this bounded search.

    Dependent claim: The FCC entered a final forfeiture and later reported referral for collection.

    Prohibited inference: Do not call the penalty paid, collected, reversed, judicially affirmed, or finally resolved without an attributable disposition.

    Queries and repositories

    Queries

    • site:fcc.gov Steve Kramer 6 million forfeiture collection DOJ
    • site:justice.gov Steve Kramer FCC robocall forfeiture collection court
    • site:docs.fcc.gov annual financial report Steve Kramer forfeiture referred DOJ

    Repositories

    • FCC orders and financial reports; DOJ and public court records

    Correction trigger: A primary, independently attributable, adjudicative, or corrected record materially changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, implementation record, independent field study, correction, or affected-person record appears.

  21. v3Meta ayahuasca downstream restoration and ranking repairCONFIRMS
    Search ID
    SEARCH-21-META-AYAHUASCA-V3
    Prior search
    SEARCH-12-META-AYAHUASCA-V2
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: VISIBLE_RESTORATION_RETAINED_DOWNSTREAM_RANKING_STRIKE_AND_REACH_REPAIR_NOT_RETRIEVED

    Known blind spots: Internal ranking, feature, account-state, and recommendation systems are not publicly observable.

    Dependent claim: The post was restored after an appellate decision.

    Prohibited inference: Do not equate visible restoration with correction of strikes, recommendation eligibility, reach, monetization, or complete repair.

    Queries and repositories

    Queries

    • site:oversightboard.com ayahuasca implementation ranking strike restoration
    • site:transparency.meta.com oversight board ayahuasca implementation account strike

    Repositories

    • Meta Oversight Board decisions and implementation material

    Correction trigger: A primary, independently attributable, adjudicative, or corrected record materially changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, implementation record, independent field study, correction, or affected-person record appears.

  22. v3Robodebt compensation, scheme operation, data correction, and repeated-error preventionSUPPLEMENTS
    Search ID
    SEARCH-22-ROBODEBT-V3
    Prior search
    SEARCH-13-ROBODEBT-V2
    Search date
    2026-07-28
    Public wording changed
    YES

    Result: 2026_SETTLEMENT_AND_RESOLUTION_SCHEME_OPERATION_DOCUMENTED_COMPLETE_DATA_REPAIR_AND_REPEAT_PREVENTION_NOT_ESTABLISHED

    Known blind spots: Individual outcomes, exclusions, copied records, downstream systems, and long-term repeat-error metrics remain incompletely public.

    Dependent claim: Additional compensation and resolution mechanisms are operational or approved.

    Prohibited inference: Do not call settlement, scheme access, or corruption findings complete repair for every affected person or system.

    Queries and repositories

    Queries

    • site:servicesaustralia.gov.au 2026 Robodebt class action settlement income apportionment resolution scheme
    • site:nacc.gov.au Operation Myrtleford Robodebt report
    • site:ag.gov.au Robodebt automated decision reforms data correction

    Repositories

    • Services Australia, Federal Court notices, NACC, Attorney-General and Royal Commission records

    Correction trigger: A primary, independently attributable, adjudicative, or corrected record materially changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, implementation record, independent field study, correction, or affected-person record appears.

  23. v3C2PA durable credentials and field performanceCONFIRMS
    Search ID
    SEARCH-23-C2PA-V3
    Prior search
    SEARCH-14-C2PA-V2
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: CURRENT_SPECIFICATION_AND_LAYERED_RECOVERY_GUIDANCE_RETRIEVED_HIGH_STAKES_CROSS_PLATFORM_FIELD_PERFORMANCE_NOT_RETRIEVED

    Known blind spots: Real platform implementations, screenshots, re-encoding, copied assets, and user-interface presentation vary.

    Dependent claim: C2PA provides signed provenance with durability and recovery approaches.

    Prohibited inference: Do not treat credential presence as truth or credential absence as falsity.

    Queries and repositories

    Queries

    • C2PA 2.4 durable content credentials field performance re-encoding screenshots 2026
    • independent C2PA implementation evaluation metadata stripping recovery

    Repositories

    • C2PA specification and explainer; independent security and deployment research

    Correction trigger: A primary, independently attributable, adjudicative, or corrected record materially changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, implementation record, independent field study, correction, or affected-person record appears.

  24. v3Production-grade machine-unlearning auditabilitySUPPLEMENTS
    Search ID
    SEARCH-24-UNLEARNING-V3
    Prior search
    SEARCH-15-UNLEARNING-V2
    Search date
    2026-07-28
    Public wording changed
    YES

    Result: BLACK_BOX_AUDIT_FRAMEWORK_RETRIEVED_PRODUCTION_EXACT_UNLEARNING_EQUIVALENCE_NOT_RETRIEVED

    Known blind spots: Production models, training data, parameters, and retraining references are generally proprietary.

    Dependent claim: Black-box statistical auditing can test whether an unlearning claim is distinguishable from a retraining reference under bounded conditions.

    Prohibited inference: Do not describe source deletion, refusal, model editing, or an unaudited vendor claim as exact unlearning.

    Queries and repositories

    Queries

    • Google Research new framework auditing machine unlearning June 2026
    • production foundation model unlearning external audit exact retraining reference 2026

    Repositories

    • Google Research, NIST, primary research, independent audits

    Correction trigger: A primary, independently attributable, adjudicative, or corrected record materially changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, implementation record, independent field study, correction, or affected-person record appears.

  25. v3Mental-state inference field validationSUPPLEMENTS
    Search ID
    SEARCH-25-MENTAL-STATE-V3
    Prior search
    SEARCH-16-MENTAL-STATE-V2
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: ADDITIONAL_DEPLOYMENT_RESEARCH_RETRIEVED_CONSTRUCT_SPECIFIC_CONSEQUENTIAL_FIELD_VALIDITY_NOT_RETRIEVED

    Known blind spots: Proprietary models, unstable ground truth, changing versions, and undisclosed action thresholds limit evaluation.

    Dependent claim: Systems continue to be studied and deployed for proxy-signal classification.

    Prohibited inference: Do not convert facial, vocal, gaze, neural, or interaction features into psychological fact without construct-valid evidence.

    Queries and repositories

    Queries

    • 2026 ACM emotion recognition agile meetings field validity
    • independent workplace emotion AI construct validity consequential decisions 2026
    • job interview emotion AI field validation subgroup calibration

    Repositories

    • Peer-reviewed HCI and psychology; EU law; regulator and system-specific records

    Correction trigger: A primary, independently attributable, adjudicative, or corrected record materially changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, implementation record, independent field study, correction, or affected-person record appears.

  26. v3Ukraine surrender deepfake recipient-level and strategic effectsCONFIRMS
    Search ID
    SEARCH-26-UKRAINE-DEEPFAKE-V3
    Prior search
    SEARCH-17-UKRAINE-DEEPFAKE-V2
    Search date
    2026-07-28
    Public wording changed
    NO

    Result: DISCOURSE_AND_TRUST_THEMATIC_RESEARCH_RETRIEVED_RECIPIENT_LEVEL_BELIEF_BEHAVIOR_AND_FRONTLINE_EFFECT_NOT_RETRIEVED

    Known blind spots: Military communications, recipient surveys, platform exposure data, and classified assessments may be unavailable.

    Dependent claim: The artifact, hacked distribution context, rapid correction, and public discourse are documented.

    Prohibited inference: Do not infer recipient belief, troop behavior, operational disruption, or strategic effect from tweets, availability, or media coverage.

    Queries and repositories

    Queries

    • Zelensky surrender deepfake recipient survey belief behavior frontline effect
    • Ukraine deepfake trust PLOS ONE thematic tweets recipient exposure
    • Ukraine 24 hacked broadcast deepfake operational impact

    Repositories

    • Peer-reviewed research, Ukrainian records, DFRLab, platform and military records

    Correction trigger: A primary, independently attributable, adjudicative, or corrected record materially changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, implementation record, independent field study, correction, or affected-person record appears.

  27. v3VIGINUM false positives, remedy, oversight, and affected-community evidenceSUPPLEMENTS
    Search ID
    SEARCH-27-VIGINUM-V3
    Prior search
    SEARCH-18-VIGINUM-V2
    Search date
    2026-07-28
    Public wording changed
    YES

    Result: 2026_MISSION_AND_MUNICIPAL_ACTIVITY_CURRENTNESS_RETRIEVED_PUBLIC_FALSE_POSITIVE_REMEDY_AND_AFFECTED_COMMUNITY_OUTCOME_NOT_RETRIEVED

    Known blind spots: Internal logs, private submissions, nonpublic corrections, and unindexed affected-community records remain outside the search.

    Dependent claim: The current mission page and municipal report document remit and activity.

    Prohibited inference: Do not infer independent accuracy, absence of false positives, complete oversight, remedy, or community acceptance from operator publications.

    Queries and repositories

    Queries

    • site:sgdsn.gouv.fr VIGINUM 2026 municipal elections four operations decree 2026-70
    • VIGINUM false positive correction appeal affected community oversight 2026
    • site:sgdsn.gouv.fr VIGINUM committee ethics remedy false positive

    Repositories

    • SGDSN/VIGINUM current mission page and reports; ethics, legal, oversight, journalism, and affected-community records

    Correction trigger: A primary, independently attributable, adjudicative, or corrected record materially changes the stated status.

    Reopening trigger: Repeat when a final report, audit, court disposition, implementation record, independent field study, correction, or affected-person record appears.

  28. v4FTC companion-chatbot inquiry outcomeCONFIRMS
    Search ID
    SEARCH-28-FTC-COMPANION-V4
    Prior search
    SEARCH-19-FTC-COMPANION-V3
    Search date
    2026-07-29
    Public wording changed
    NO

    Result: OPEN_INQUIRY_RECONFIRMED_NO_PUBLIC_FINAL_FINDINGS_OR_CLOSURE_RETRIEVED_AS_OF_2026-07-29

    Known blind spots: Private records, unpublished datasets, internal audits, recipient-level evidence, attachments, downstream vendor records, and later proceedings may remain unavailable.

    Dependent claim: FTC companion-chatbot inquiry outcome

    Prohibited inference: Do not infer nonexistence, destruction, universal validity, complete repair, recipient belief, behavior change, strategic effect, or legal closure from this bounded search.

    Queries and repositories

    Queries

    • site:ftc.gov companion chatbots inquiry outcome 2026 FTC
    • site:ftc.gov 6(b) AI chatbots acting as companions final report

    Repositories

    • Official issuing-body websites
    • Official legal or regulatory repositories
    • Primary technical specification or research repositories

    Correction trigger: A direct official, judicial, independently validated, or attributable affected-person record materially changes the public wording.

    Reopening trigger: A final order, implementation report, appeal, audit, field-validation study, recipient study, remedy outcome, or oversight finding is published.

  29. v4FCC Biden robocall collection, payment, review, and court statusCONFIRMS
    Search ID
    SEARCH-29-FCC-ROBOCALL-V4
    Prior search
    SEARCH-20-FCC-ROBOCALL-V3
    Search date
    2026-07-29
    Public wording changed
    NO

    Result: FINAL_FORFEITURE_AND_2025_REPORT_RETRIEVED_PAYMENT_COLLECTION_OR_COURT_DISPOSITION_NOT_RETRIEVED

    Known blind spots: Private records, unpublished datasets, internal audits, recipient-level evidence, attachments, downstream vendor records, and later proceedings may remain unavailable.

    Dependent claim: FCC Biden robocall collection, payment, review, and court status

    Prohibited inference: Do not infer nonexistence, destruction, universal validity, complete repair, recipient belief, behavior change, strategic effect, or legal closure from this bounded search.

    Queries and repositories

    Queries

    • site:fcc.gov Steve Kramer Biden robocall 2026 DOJ collection payment court
    • site:justice.gov Steve Kramer FCC robocall collection

    Repositories

    • Official issuing-body websites
    • Official legal or regulatory repositories
    • Primary technical specification or research repositories

    Correction trigger: A direct official, judicial, independently validated, or attributable affected-person record materially changes the public wording.

    Reopening trigger: A final order, implementation report, appeal, audit, field-validation study, recipient study, remedy outcome, or oversight finding is published.

  30. v4Meta ayahuasca restoration and downstream repairSUPPLEMENTS
    Search ID
    SEARCH-30-META-AYAHUASCA-V4
    Prior search
    SEARCH-21-META-AYAHUASCA-V3
    Search date
    2026-07-29
    Public wording changed
    NO

    Result: RESTORATION_AND_PARTIAL_POLICY_IMPLEMENTATION_RETRIEVED_COMPLETE_RANKING_STRIKE_REACH_AND_INCOME_REPAIR_NOT_RETRIEVED

    Known blind spots: Private records, unpublished datasets, internal audits, recipient-level evidence, attachments, downstream vendor records, and later proceedings may remain unavailable.

    Dependent claim: Meta ayahuasca restoration and downstream repair

    Prohibited inference: Do not infer nonexistence, destruction, universal validity, complete repair, recipient belief, behavior change, strategic effect, or legal closure from this bounded search.

    Queries and repositories

    Queries

    • site:oversightboard.com ayahuasca restoration implementation ranking strike reach
    • site:oversightboard.com Ayahuasca brew recommendation implementation

    Repositories

    • Official issuing-body websites
    • Official legal or regulatory repositories
    • Primary technical specification or research repositories

    Correction trigger: A direct official, judicial, independently validated, or attributable affected-person record materially changes the public wording.

    Reopening trigger: A final order, implementation report, appeal, audit, field-validation study, recipient study, remedy outcome, or oversight finding is published.

  31. v4Robodebt compensation, data correction, implementation, and repeat-error preventionSUPPLEMENTS
    Search ID
    SEARCH-31-ROBODEBT-V4
    Prior search
    SEARCH-22-ROBODEBT-V3
    Search date
    2026-07-29
    Public wording changed
    YES

    Result: FEDERAL_COURT_APPROVED_2026_ADDITIONAL_SETTLEMENT_AND_IMPLEMENTATION_CONTEXT_RETRIEVED_COMPLETE_DATA_REPAIR_AND_REPEAT_PREVENTION_NOT_ESTABLISHED

    Known blind spots: Private records, unpublished datasets, internal audits, recipient-level evidence, attachments, downstream vendor records, and later proceedings may remain unavailable.

    Dependent claim: Robodebt compensation, data correction, implementation, and repeat-error prevention

    Prohibited inference: Do not infer nonexistence, destruction, universal validity, complete repair, recipient belief, behavior change, strategic effect, or legal closure from this bounded search.

    Queries and repositories

    Queries

    • site:servicesaustralia.gov.au new robodebt class action settlement 2026
    • site:servicesaustralia.gov.au robodebt implementation recommendations repeat prevention

    Repositories

    • Official issuing-body websites
    • Official legal or regulatory repositories
    • Primary technical specification or research repositories

    Correction trigger: A direct official, judicial, independently validated, or attributable affected-person record materially changes the public wording.

    Reopening trigger: A final order, implementation report, appeal, audit, field-validation study, recipient study, remedy outcome, or oversight finding is published.

  32. v4C2PA durability, trust-list governance, and recoverySUPPLEMENTS
    Search ID
    SEARCH-32-C2PA-V4
    Prior search
    SEARCH-23-C2PA-V3
    Search date
    2026-07-29
    Public wording changed
    YES

    Result: C2PA_2_4_DURABLE_CREDENTIAL_SOFT_BINDING_AND_TRUST_LIST_GUIDANCE_RETRIEVED_TRUTH_AND_FIELD_SURVIVAL_NOT_ESTABLISHED

    Known blind spots: Private records, unpublished datasets, internal audits, recipient-level evidence, attachments, downstream vendor records, and later proceedings may remain unavailable.

    Dependent claim: C2PA durability, trust-list governance, and recovery

    Prohibited inference: Do not infer nonexistence, destruction, universal validity, complete repair, recipient belief, behavior change, strategic effect, or legal closure from this bounded search.

    Queries and repositories

    Queries

    • site:c2pa.org specification 2.4 durable content credentials trust list soft binding
    • site:c2pa.org Content Credentials absence not proof false

    Repositories

    • Official issuing-body websites
    • Official legal or regulatory repositories
    • Primary technical specification or research repositories

    Correction trigger: A direct official, judicial, independently validated, or attributable affected-person record materially changes the public wording.

    Reopening trigger: A final order, implementation report, appeal, audit, field-validation study, recipient study, remedy outcome, or oversight finding is published.

  33. v4Machine-unlearning exactness, auditability, production deployment, and legal implementationSUPPLEMENTS
    Search ID
    SEARCH-33-UNLEARNING-V4
    Prior search
    SEARCH-24-UNLEARNING-V3
    Search date
    2026-07-29
    Public wording changed
    YES

    Result: CURRENT_RESEARCH_RECONFIRMS_EXACT_RETRAINING_REFERENCE_AND_OPEN_AUDITABILITY_PRODUCTION_AND_LEGAL_IMPLEMENTATION_GAPS

    Known blind spots: Private records, unpublished datasets, internal audits, recipient-level evidence, attachments, downstream vendor records, and later proceedings may remain unavailable.

    Dependent claim: Machine-unlearning exactness, auditability, production deployment, and legal implementation

    Prohibited inference: Do not infer nonexistence, destruction, universal validity, complete repair, recipient belief, behavior change, strategic effect, or legal closure from this bounded search.

    Queries and repositories

    Queries

    • site:nist.gov machine unlearning exact approximate
    • site:arxiv.org machine unlearning auditing 2026 exact retraining benchmark

    Repositories

    • Official issuing-body websites
    • Official legal or regulatory repositories
    • Primary technical specification or research repositories

    Correction trigger: A direct official, judicial, independently validated, or attributable affected-person record materially changes the public wording.

    Reopening trigger: A final order, implementation report, appeal, audit, field-validation study, recipient study, remedy outcome, or oversight finding is published.

  34. v4Mental-state inference construct and field validitySUPPLEMENTS
    Search ID
    SEARCH-34-MENTAL-STATE-V4
    Prior search
    SEARCH-25-MENTAL-STATE-V3
    Search date
    2026-07-29
    Public wording changed
    NO

    Result: CURRENT_EU_GUIDANCE_RETRIEVED_PROHIBITION_SCOPE_DOES_NOT_VALIDATE_CONSTRUCT_SPECIFIC_FIELD_INFERENCE

    Known blind spots: Private records, unpublished datasets, internal audits, recipient-level evidence, attachments, downstream vendor records, and later proceedings may remain unavailable.

    Dependent claim: Mental-state inference construct and field validity

    Prohibited inference: Do not infer nonexistence, destruction, universal validity, complete repair, recipient belief, behavior change, strategic effect, or legal closure from this bounded search.

    Queries and repositories

    Queries

    • site:ec.europa.eu AI Act emotion recognition workplace education guidance
    • mental state inference construct validity field deployment 2026

    Repositories

    • Official issuing-body websites
    • Official legal or regulatory repositories
    • Primary technical specification or research repositories

    Correction trigger: A direct official, judicial, independently validated, or attributable affected-person record materially changes the public wording.

    Reopening trigger: A final order, implementation report, appeal, audit, field-validation study, recipient study, remedy outcome, or oversight finding is published.

  35. v4Ukraine surrender-deepfake recipient-level effectsSUPPLEMENTS
    Search ID
    SEARCH-35-UKRAINE-DEEPFAKE-V4
    Prior search
    SEARCH-26-UKRAINE-DEEPFAKE-V3
    Search date
    2026-07-29
    Public wording changed
    NO

    Result: OFFICIAL_ANALYTICAL_REJECTION_AND_RAPID_REFUTATION_CONTEXT_RETRIEVED_INDEPENDENT_RECIPIENT_BELIEF_BEHAVIOR_AND_STRATEGIC_EFFECT_NOT_RETRIEVED

    Known blind spots: Private records, unpublished datasets, internal audits, recipient-level evidence, attachments, downstream vendor records, and later proceedings may remain unavailable.

    Dependent claim: Ukraine surrender-deepfake recipient-level effects

    Prohibited inference: Do not infer nonexistence, destruction, universal validity, complete repair, recipient belief, behavior change, strategic effect, or legal closure from this bounded search.

    Queries and repositories

    Queries

    • site:cpd.gov.ua Zelensky surrender deepfake target audience effect
    • Ukraine surrender deepfake recipient belief behavior frontline effect official

    Repositories

    • Official issuing-body websites
    • Official legal or regulatory repositories
    • Primary technical specification or research repositories

    Correction trigger: A direct official, judicial, independently validated, or attributable affected-person record materially changes the public wording.

    Reopening trigger: A final order, implementation report, appeal, audit, field-validation study, recipient study, remedy outcome, or oversight finding is published.

  36. v4VIGINUM false positives, correction, remedy, oversight, and affected-community evidenceSUPPLEMENTS
    Search ID
    SEARCH-36-VIGINUM-V4
    Prior search
    SEARCH-27-VIGINUM-V3
    Search date
    2026-07-29
    Public wording changed
    YES

    Result: 2026_ACTIVITY_OVERSIGHT_AND_INDIRECT_DATA_RIGHTS_CONTEXT_RETRIEVED_PUBLIC_FALSE_POSITIVE_REMEDY_AND_AFFECTED_COMMUNITY_OUTCOME_REGISTER_NOT_RETRIEVED

    Known blind spots: Private records, unpublished datasets, internal audits, recipient-level evidence, attachments, downstream vendor records, and later proceedings may remain unavailable.

    Dependent claim: VIGINUM false positives, correction, remedy, oversight, and affected-community evidence

    Prohibited inference: Do not infer nonexistence, destruction, universal validity, complete repair, recipient belief, behavior change, strategic effect, or legal closure from this bounded search.

    Queries and repositories

    Queries

    • site:sgdsn.gouv.fr VIGINUM committee ethics scientific report remedy false positives
    • site:legifrance.gouv.fr VIGINUM access rectification erasure CNIL

    Repositories

    • Official issuing-body websites
    • Official legal or regulatory repositories
    • Primary technical specification or research repositories

    Correction trigger: A direct official, judicial, independently validated, or attributable affected-person record materially changes the public wording.

    Reopening trigger: A final order, implementation report, appeal, audit, field-validation study, recipient study, remedy outcome, or oversight finding is published.

  37. v5Bounded search
    Search ID
    SEARCH-37-FTC-COMPANION-V5
    Prior search
    NONE
    Search date
    2026-07-29
    Public wording changed
    NO

    Result:

    Known blind spots:

    Dependent claim: WIP.62 currentness and source-quality wording only.

    Prohibited inference: Do not treat inquiry as final finding, liability, causation, or prevalence.

    Queries and repositories

    Queries

      Repositories

        Correction trigger: The issuing body publishes a superseding, corrected, final, or materially different record.

        Reopening trigger: A final report, implementation decision, adjudication, independent evaluation, affected-person evidence, or contrary primary source becomes available.

      • v5Bounded search
        Search ID
        SEARCH-38-C2PA-V5
        Prior search
        NONE
        Search date
        2026-07-29
        Public wording changed
        NO

        Result:

        Known blind spots:

        Dependent claim: WIP.62 currentness and source-quality wording only.

        Prohibited inference: Do not treat provenance as proof that depicted content is true or credentials survive every transformation.

        Queries and repositories

        Queries

          Repositories

            Correction trigger: The issuing body publishes a superseding, corrected, final, or materially different record.

            Reopening trigger: A final report, implementation decision, adjudication, independent evaluation, affected-person evidence, or contrary primary source becomes available.

          • v5Bounded search
            Search ID
            SEARCH-39-EU-AI-ACT-V5
            Prior search
            NONE
            Search date
            2026-07-29
            Public wording changed
            NO

            Result:

            Known blind spots:

            Dependent claim: WIP.62 currentness and source-quality wording only.

            Prohibited inference: Do not universalize EU law or claim uniform enforcement outcomes.

            Queries and repositories

            Queries

              Repositories

                Correction trigger: The issuing body publishes a superseding, corrected, final, or materially different record.

                Reopening trigger: A final report, implementation decision, adjudication, independent evaluation, affected-person evidence, or contrary primary source becomes available.

              • v5Bounded search
                Search ID
                SEARCH-40-COE-CETS225-V5
                Prior search
                NONE
                Search date
                2026-07-29
                Public wording changed
                NO

                Result:

                Known blind spots:

                Dependent claim: WIP.62 currentness and source-quality wording only.

                Prohibited inference: Do not infer domestic implementation or entry into force beyond the exact official status record.

                Queries and repositories

                Queries

                  Repositories

                    Correction trigger: The issuing body publishes a superseding, corrected, final, or materially different record.

                    Reopening trigger: A final report, implementation decision, adjudication, independent evaluation, affected-person evidence, or contrary primary source becomes available.

                  • v5Bounded search
                    Search ID
                    SEARCH-41-UNESCO-NEUROTECH-V5
                    Prior search
                    NONE
                    Search date
                    2026-07-29
                    Public wording changed
                    NO

                    Result:

                    Known blind spots:

                    Dependent claim: WIP.62 currentness and source-quality wording only.

                    Prohibited inference: Do not describe the non-binding recommendation as universal binding law.

                    Queries and repositories

                    Queries

                      Repositories

                        Correction trigger: The issuing body publishes a superseding, corrected, final, or materially different record.

                        Reopening trigger: A final report, implementation decision, adjudication, independent evaluation, affected-person evidence, or contrary primary source becomes available.

                      • v5Bounded search
                        Search ID
                        SEARCH-42-VIGINUM-V5
                        Prior search
                        NONE
                        Search date
                        2026-07-29
                        Public wording changed
                        NO

                        Result:

                        Known blind spots:

                        Dependent claim: WIP.62 currentness and source-quality wording only.

                        Prohibited inference: Do not infer false-positive rates, remedy outcomes, or universal democratic legitimacy without independent evidence.

                        Queries and repositories

                        Queries

                          Repositories

                            Correction trigger: The issuing body publishes a superseding, corrected, final, or materially different record.

                            Reopening trigger: A final report, implementation decision, adjudication, independent evaluation, affected-person evidence, or contrary primary source becomes available.

                          • v5Bounded search
                            Search ID
                            SEARCH-43-DEEPFAKE-AUTH-V5
                            Prior search
                            NONE
                            Search date
                            2026-07-29
                            Public wording changed
                            NO

                            Result:

                            Known blind spots:

                            Dependent claim: WIP.62 currentness and source-quality wording only.

                            Prohibited inference: Do not treat authentication tooling as measured audience correction or behavior change.

                            Queries and repositories

                            Queries

                              Repositories

                                Correction trigger: The issuing body publishes a superseding, corrected, final, or materially different record.

                                Reopening trigger: A final report, implementation decision, adjudication, independent evaluation, affected-person evidence, or contrary primary source becomes available.

                              • v5Bounded search
                                Search ID
                                SEARCH-44-MENTAL-INFERENCE-V5
                                Prior search
                                NONE
                                Search date
                                2026-07-29
                                Public wording changed
                                NO

                                Result:

                                Known blind spots:

                                Dependent claim: WIP.62 currentness and source-quality wording only.

                                Prohibited inference: Do not turn legal restriction into proof of a universal scientific proposition.

                                Queries and repositories

                                Queries

                                  Repositories

                                    Correction trigger: The issuing body publishes a superseding, corrected, final, or materially different record.

                                    Reopening trigger: A final report, implementation decision, adjudication, independent evaluation, affected-person evidence, or contrary primary source becomes available.

                                  • v5Bounded search
                                    Search ID
                                    SEARCH-45-REVIEW-IMPORT-STANDARDS-V5
                                    Prior search
                                    NONE
                                    Search date
                                    2026-07-29
                                    Public wording changed
                                    NO

                                    Result:

                                    Known blind spots:

                                    Dependent claim: WIP.62 currentness and source-quality wording only.

                                    Prohibited inference: Do not claim RFC canonicalization alone provides signatures, identity, or adjudication.

                                    Queries and repositories

                                    Queries

                                      Repositories

                                        Correction trigger: The issuing body publishes a superseding, corrected, final, or materially different record.

                                        Reopening trigger: A final report, implementation decision, adjudication, independent evaluation, affected-person evidence, or contrary primary source becomes available.

                                      REMEDY AND SCIENTIFIC VALIDITY

                                      Two chains that prevent paper compliance and scientific overclaiming

                                      A nominal appeal is not complete repair. A high classification score is not proof that a system measures the claimed psychological construct.

                                      Thirteen-stage remedy chain

                                      1. 01Notice
                                      2. 02Comprehension
                                      3. 03Access
                                      4. 04Relevant evidence
                                      5. 05Authoritative review
                                      6. 06Reversal
                                      7. 07Immediate restoration
                                      8. 08Source-data correction
                                      9. 09Inferred-data correction
                                      10. 10Downstream propagation
                                      11. 11Compensation
                                      12. 12Repeat-error prevention
                                      13. 13Unresolved harm

                                      Sixteen-stage validity chain

                                      1. 01Sensor or input quality
                                      2. 02Feature extraction
                                      3. 03Label quality
                                      4. 04Construct validity
                                      5. 05Criterion validity
                                      6. 06Calibration
                                      7. 07Generalization
                                      8. 08Base-rate sensitivity
                                      9. 09False-positive burden
                                      10. 10False-negative burden
                                      11. 11Subgroup performance
                                      12. 12Human override
                                      13. 13Automation bias
                                      14. 14Institutional action
                                      15. 15Downstream consequence
                                      16. 16Remedy

                                      REPOSITORY-AUTHORED BOUNDED RESEARCH

                                      Seven predictive deployments with outcome, validity, and remedy fields

                                      No dedicated owner-supplied predictive-deployment report was present in the WIP.55 intake. WIP.62 therefore keeps the synthesis visibly repository-authored while versioning each deployment review, recording longitudinal changes and currentness relations, and retaining independent outcome evidence, legal and model context, evaluation periods, base-rate and error gaps, institutional action, feedback loops, notice, correction, deletion, and affected-person evidence.

                                      Aggregate forecasting

                                      Examples: Influenza activity, forced displacement, conflict risk

                                      Usually planning-oriented; uncertainty, resource allocation, omission, and politicized interpretation remain material.

                                      Case routing and prioritization

                                      Examples: Visa application complexity routing

                                      Not necessarily a final decision, but can alter delay, scrutiny, workload, and automation bias.

                                      Person or family risk scoring

                                      Examples: Police risk lists, child-welfare screening, fraud-risk indications

                                      Heightened notice, validation, subgroup burden, feedback-loop, human-authority, appeal, and downstream-correction requirements.

                                      20Outcome-evidence dimensionsFIELDS REVIEWED PER DEPLOYMENT
                                      1. 01Legal authority
                                      2. 02Model or rules-engine version
                                      3. 03Unit of prediction
                                      4. 04Target event and horizon
                                      5. 05Data sources and exclusions
                                      6. 06Training and evaluation period
                                      7. 07Calibration and out-of-sample performance
                                      8. 08Base rate
                                      9. 09False-positive burden
                                      10. 10False-negative burden
                                      11. 11Subgroup performance and missingness
                                      12. 12Human reviewer authority
                                      13. 13Automation bias
                                      14. 14Institutional action
                                      15. 15Feedback loop
                                      16. 16Notice and appeal
                                      17. 17Correction and deletion
                                      18. 18Downstream repair
                                      19. 19Affected-person or community evidence
                                      20. 20Current status and reopening trigger
                                      Deployment purpose, system class, unit, institutional action, outcome status, and unresolved remedy
                                      DeploymentCurrentness relationSystem classUnitInstitutional actionOutcome statusUnresolved evidence
                                      Chicago Strategic Subjects List and Crime and Victimization Risk ModelPolicing and public safety UNCHANGEDNO_SUBSTANTIVE_CHANGE PERSON_LEVEL_RISK_SCORING Named individuals Referral and prioritization for police-led intervention and monitoring. NO_SIGNIFICANT_PROTECTIVE_EFFECT_REPORTED_DECOMMISSIONED

                                      Were all person-level scores, tiers, exports, and derivative labels deleted or corrected after decommissioning?

                                      Public evidence of individualized notice, appeal, score correction, retained-copy deletion, or complete downstream repair was NOT_RETRIEVED.
                                      System Risk Indication (SyRI)Public-benefits and fraud risk UNCHANGEDNO_SUBSTANTIVE_CHANGE PERSON_OR_HOUSEHOLD_RISK_ANALYSIS People and households within selected areas or investigations Risk notifications could prompt further human review or investigation. JUDICIALLY_INVALIDATED_DATA_DESTRUCTION_OFFICIALLY_REPORTED

                                      Did destruction encompass every derivative, backup, audit, and downstream record, and what community-level repair followed?

                                      The government states all SyRI-processed data were destroyed; independent forensic verification, backup coverage, and individualized notice or remedy were NOT_RETRIEVED.
                                      Allegheny Family Screening ToolChild welfare intake SUPPLEMENTSCURRENTNESS_REFRESH FAMILY_LEVEL_RISK_SCREENING Children and families named in hotline referrals Screening for investigation and assignment to further child-welfare processing. MIXED_OUTCOMES_AND_DISPARITY_FINDINGS_CAUSAL_ATTRIBUTION_LIMITED

                                      Which observed changes are attributable to the model rather than policy, workforce, referral mix, or pandemic-era changes?

                                      Public evidence of family notice about score use, score correction, deletion, appeal, or downstream repair was NOT_RETRIEVED.
                                      UK Home Office Complexity Application Routing Solution — VisitsImmigration workflow SUPPLEMENTSCURRENTNESS_REFRESH CASE_ROUTING_AND_PRIORITIZATION Visitor visa applications Assignment to workflow and staff skill level; complex applications may receive additional checks and take longer. PERMANENT_WORKFLOW_ROUTER_INSPECTION_FOUND_GENERAL_ACCURACY_AND_ASSURANCE_GAPS

                                      What are current error rates, delay effects, nationality impacts, and applicant remedy outcomes under the production version?

                                      Applicants are not described as receiving individualized routing explanations or a distinct CARS appeal; visa decision remedies are separate. Routing-data correction and downstream repair were NOT_RETRIEVED.
                                      CDC FluSightPublic-health planning SUPPLEMENTSSEASON_STATUS_REFRESH AGGREGATE_POPULATION_FORECASTING Population-level influenza activity over time Planning, situational awareness, and messaging to health-care and public-health partners. ACTIVE_TRANSPARENT_SEASONAL_EVALUATION_WITH_DOCUMENTED_FAILURES

                                      How often do forecasts change allocation or messaging, and what measurable benefit or harm follows?

                                      Public forecasts and evaluations are transparent; individual appeal and deletion are not applicable to aggregate outputs, but correction and version history remain important.
                                      UNHCR forced-displacement forecastingHumanitarian planning UNCHANGEDNO_SUBSTANTIVE_CHANGE AGGREGATE_DISPLACEMENT_FORECASTING Aggregate displacement flows and humanitarian demand Scenario planning, preparedness, supply-chain positioning, and operational prioritization. HISTORICAL_AND_EMERGING_OPERATIONAL_USE_INDEPENDENT_IMPACT_NOT_RETRIEVED

                                      Did forecast use measurably improve assistance, reduce delay, or create geographic or demographic omission?

                                      Aggregate forecasts do not create a direct individual appeal path; model governance, correction, and community participation remain relevant.
                                      European Commission JRC Dynamic Conflict Risk ModelConflict early warning UNCHANGEDNO_SUBSTANTIVE_CHANGE AGGREGATE_STRUCTURAL_RISK_FORECASTING Country- or area-level conflict risk over defined horizons Analytical prioritization, horizon scanning, and potential early-warning follow-up. ACTIVE_ANALYTICAL_EARLY_WARNING_INTERNAL_PERFORMANCE_RESULTS_NOT_PUBLIC

                                      What are threshold-specific false alarm and miss rates, regional calibration, and measurable policy outcomes?

                                      No individual score is published, but countries and communities may be affected by prioritization; public correction, challenge, and affected-community mechanisms were NOT_RETRIEVED.
                                      Chicago Strategic Subjects List and Crime and Victimization Risk Model — full outcome and evidence boundary
                                      Stable ID
                                      PRED-01-CHICAGO-SSL-CVRM
                                      Jurisdiction
                                      Chicago, Illinois, United States
                                      Operator
                                      Chicago Police Department
                                      Current status
                                      DECOMMISSIONED_2019-11-01
                                      Prior review
                                      2026-07-28
                                      Current review
                                      2026-07-28
                                      Source version
                                      WIP58_LONGITUDINAL_REFRESH_V2
                                      Currentness relation
                                      UNCHANGED
                                      Validation status
                                      OFFICIAL_OVERSIGHT_DOCUMENTED_INDEPENDENT_CAUSAL_BENEFIT_NOT_ESTABLISHED
                                      Specialist disposition
                                      NOT_COMPLETED

                                      Legal authority

                                      Police program authority and federal grant context are described in OIG/NIJ records; a claim of separate judicial authorization is not made.

                                      Model or rules version

                                      2013 Strategic Subjects List pilot; later CVRM / person-threat-victimization risk tiers. Exact production version history and weights are not fully public in the cited records.

                                      Target and time horizon

                                      Target: Involvement in gun violence as victim or alleged offender.

                                      Horizon: The public evaluation and advisory do not provide one stable, universally applicable horizon across SSL and CVRM versions.

                                      Data and exclusions

                                      Historical police and network data described in the source record; complete feature lineage, exclusions, missingness handling, and retained copies are not fully public.

                                      Training and evaluation

                                      SSL pilot implemented in 2013; quasi-experimental evaluation compared listed people and city-level trends. Later CVRM governance reviewed through decommissioning in 2019.

                                      Calibration and external performance

                                      NIJ reports no significant effect on homicide or shooting victimization attributable to list placement; listed people were more likely to be arrested for a shooting. Calibration curves and external-site validation were not retrieved.

                                      Base rate and error burden

                                      Base rate: NOT_RETRIEVED for the deployed threshold and target event.

                                      False positives: NOT_RETRIEVED at the person-level operating threshold; the absence of benefit plus elevated contact/arrest risk makes this a material unresolved burden.

                                      False negatives: NOT_RETRIEVED.

                                      Subgroups and missingness

                                      OIG identified governance and data-quality concerns; complete subgroup calibration and missing-data results were not retrieved.

                                      Human authority and automation bias

                                      Authority: Police commanders and intervention staff retained action authority; the score structured who received attention.

                                      Risk: Risk tiering could anchor attention and produce more contact even without validated protective benefit.

                                      Institutional action and feedback loop

                                      Action: Referral and prioritization for police-led intervention and monitoring.

                                      Loop: Police contact and arrest data can become future model inputs, reinforcing enforcement patterns.

                                      Notice, appeal, correction, deletion

                                      Public evidence of individualized notice, appeal, score correction, retained-copy deletion, or complete downstream repair was NOT_RETRIEVED.

                                      Affected-person or community evidence

                                      No project-linked representative affected-person study was retrieved in the bounded WIP.57 search.

                                      Independent outcome evidence

                                      Quasi-experimental NIJ evaluation plus Chicago OIG decommissioning advisory.

                                      Unresolved question

                                      Were all person-level scores, tiers, exports, and derivative labels deleted or corrected after decommissioning?

                                      Source supports: The OIG documents the models, governance concerns, and decommissioning.

                                      Source does not establish: Crime reduction caused by the models, complete downstream repair, or universal properties of predictive policing.

                                      Primary or authoritative records

                                      Correction trigger: An official audit or dataset shows different outcome, subgroup, or deletion findings.

                                      Reopening trigger: New litigation, audit, retained-score evidence, or replacement-system record.

                                      System Risk Indication (SyRI) — full outcome and evidence boundary
                                      Stable ID
                                      PRED-02-SYRI
                                      Jurisdiction
                                      Netherlands
                                      Operator
                                      Dutch public authorities under the SyRI legal framework
                                      Current status
                                      LEGAL_FRAMEWORK_FOUND_INCOMPATIBLE_WITH_ARTICLE_8_IN_2020_JUDGMENT
                                      Prior review
                                      2026-07-28
                                      Current review
                                      2026-07-28
                                      Source version
                                      WIP58_LONGITUDINAL_REFRESH_V2
                                      Currentness relation
                                      UNCHANGED
                                      Validation status
                                      JUDICIAL_TRANSPARENCY_AND_PROPORTIONALITY_FAILURE_DOCUMENTED_PREDICTIVE_VALIDITY_NOT_ESTABLISHED
                                      Specialist disposition
                                      NOT_COMPLETED

                                      Legal authority

                                      SyRI operated under Dutch legislation and implementing rules; the District Court of The Hague held the framework incompatible with Article 8 ECHR safeguards.

                                      Model or rules version

                                      Risk-indicator matching within SyRI projects; exact algorithms, weights, and project-specific profiles were not fully public.

                                      Target and time horizon

                                      Target: Potential discrepancies or indications relevant to fraud and unlawful use of public funds.

                                      Horizon: Project-specific; NOT_RETRIEVED as a single model horizon.

                                      Data and exclusions

                                      Administrative data from participating public bodies, encrypted matching, and predetermined risk indicators. Complete project-specific feature sets, exclusions, and missingness were not retrieved.

                                      Training and evaluation

                                      Operational projects before the February 2020 judgment; a formal model-training period was not established.

                                      Calibration and external performance

                                      NOT_RETRIEVED. The judgment addressed legal safeguards, transparency, necessity, and proportionality rather than predictive calibration.

                                      Base rate and error burden

                                      Base rate: NOT_RETRIEVED.

                                      False positives: Project-specific official records discuss risk notifications, but a complete false-positive denominator was not retrieved.

                                      False negatives: NOT_RETRIEVED.

                                      Subgroups and missingness

                                      The judgment addressed neighborhood targeting and privacy risk; complete subgroup error analysis was NOT_RETRIEVED.

                                      Human authority and automation bias

                                      Authority: Responsible administrative bodies investigated risk indications; the 2022 parliamentary answer states no decisions were made based on SyRI-derived data.

                                      Risk: Opaque risk indications could focus scrutiny on residents of selected neighborhoods even without final automated decisions.

                                      Institutional action and feedback loop

                                      Action: Risk notifications could prompt further human review or investigation.

                                      Loop: Administrative investigations can generate new records that influence future fraud-control systems.

                                      Notice, appeal, correction, deletion

                                      The government states all SyRI-processed data were destroyed; independent forensic verification, backup coverage, and individualized notice or remedy were NOT_RETRIEVED.

                                      Affected-person or community evidence

                                      Litigation was brought by civil-society organizations; a representative project-linked resident outcome study was not retrieved.

                                      Independent outcome evidence

                                      Judicial invalidation plus later official destruction and no-decision-use statements.

                                      Unresolved question

                                      Did destruction encompass every derivative, backup, audit, and downstream record, and what community-level repair followed?

                                      Source supports: The court’s disposition and legal reasoning in the case.

                                      Source does not establish: That every fraud-risk system is unlawful or that every person experienced the same harm.

                                      Primary or authoritative records

                                      Correction trigger: A court, audit, or government record changes the destruction or decision-use findings.

                                      Reopening trigger: Higher-court, legislative, compensation, or implementation records.

                                      Allegheny Family Screening Tool — full outcome and evidence boundary
                                      Stable ID
                                      PRED-03-AFST
                                      Jurisdiction
                                      Allegheny County, Pennsylvania, United States
                                      Operator
                                      Allegheny County Department of Human Services
                                      Current status
                                      ACTIVE_IN_FY2026_27_OPERATOR_PLAN
                                      Prior review
                                      2026-07-28
                                      Current review
                                      2026-07-28
                                      Source version
                                      WIP58_LONGITUDINAL_REFRESH_V2
                                      Currentness relation
                                      SUPPLEMENTS
                                      Validation status
                                      OPERATOR_DOCUMENTATION_AVAILABLE_INDEPENDENT_CURRENT_VALIDITY_REVIEW_INCOMPLETE
                                      Specialist disposition
                                      NOT_COMPLETED

                                      Legal authority

                                      County child-welfare intake authority; AFST supports, rather than formally replaces, human screening decisions.

                                      Model or rules version

                                      AFST implemented in 2016 with an updated score evaluated in Phase 2; exact current production version should be verified against county methodology records.

                                      Target and time horizon

                                      Target: Likelihood of later child-welfare placement or related adverse outcomes used to support intake screening.

                                      Horizon: The evaluation includes immediate screening decisions and outcomes within 180 days.

                                      Data and exclusions

                                      County administrative records; complete current feature list and exclusion logic should be read from the version-specific methodology report. The model does not directly observe family context not present in records.

                                      Training and evaluation

                                      Pre/post implementation analyses and Phase 2 evaluation through the periods specified in the county report; COVID-19 and concurrent policy changes are noted confounders.

                                      Calibration and external performance

                                      The Phase 2 report presents adjusted risk-stratified outcomes, not a randomized trial. External-county validation was NOT_RETRIEVED.

                                      Base rate and error burden

                                      Base rate: Outcome rates are reported by referral class and risk group in the evaluation; no single universal base rate applies.

                                      False positives: A complete person-level false-positive count tied to one consequential threshold is NOT_RETRIEVED.

                                      False negatives: A complete false-negative count tied to one threshold is NOT_RETRIEVED.

                                      Subgroups and missingness

                                      Race-stratified analyses report some narrowed gaps and some context-dependent differences; exact results depend on outcome, risk group, and specification. Missingness effects remain bounded by administrative data quality.

                                      Human authority and automation bias

                                      Authority: Call screeners retain screen-in/screen-out authority and can override recommendations.

                                      Risk: The score can anchor judgment, while empirical work indicates human decisions can either reduce or reproduce algorithmic disparities.

                                      Institutional action and feedback loop

                                      Action: Screening for investigation and assignment to further child-welfare processing.

                                      Loop: Investigations and services generate future administrative records that can influence later referrals and model data.

                                      Notice, appeal, correction, deletion

                                      Public evidence of family notice about score use, score correction, deletion, appeal, or downstream repair was NOT_RETRIEVED.

                                      Affected-person or community evidence

                                      Affected-family journalism and advocacy exist, but a project-run representative family outcome study linked to the evaluation was NOT_RETRIEVED.

                                      Independent outcome evidence

                                      Operator-commissioned Phase 2 evaluation and academic human-algorithm disparity research.

                                      Unresolved question

                                      Which observed changes are attributable to the model rather than policy, workforce, referral mix, or pandemic-era changes?

                                      Source supports: Purpose, score role, and operator-described workflow.

                                      Source does not establish: Independent fairness, causal benefit, or absence of disparate burden.

                                      Primary or authoritative records

                                      Correction trigger: Independent replication or a corrected county evaluation changes outcome or disparity findings.

                                      Reopening trigger: New independent audit, litigation, model update, threshold change, or community evidence.

                                      UK Home Office Complexity Application Routing Solution — Visits — full outcome and evidence boundary
                                      Stable ID
                                      PRED-04-HOME-OFFICE-CARS
                                      Jurisdiction
                                      United Kingdom
                                      Operator
                                      UK Home Office
                                      Current status
                                      ACTIVE_ROUTING_TOOL_CONFIRMED_IN_2026_GUIDANCE_AND_RESPONSE
                                      Prior review
                                      2026-07-28
                                      Current review
                                      2026-07-28
                                      Source version
                                      WIP58_LONGITUDINAL_REFRESH_V2
                                      Currentness relation
                                      SUPPLEMENTS
                                      Validation status
                                      ROLE_AND_PURPOSE_DOCUMENTED_INDEPENDENT_GROUP_PERFORMANCE_NOT_CLOSED
                                      Specialist disposition
                                      NOT_COMPLETED

                                      Legal authority

                                      UK immigration decision-making authority and ministerial authorization for nationality-related routing profiles, as described by the inspection and transparency record.

                                      Model or rules version

                                      CARS version 3.0 was inspected; current version and profile set require verification against the current Algorithmic Transparency Recording Standard entry.

                                      Target and time horizon

                                      Target: Application complexity and routing to a decision maker with an appropriate skillset.

                                      Horizon: Workflow routing at application intake; not a long-horizon behavior prediction.

                                      Data and exclusions

                                      Application attributes, bulk-data tables, and risk/complexity indicators. Complete current profiles and missingness handling are not public because of fraud-control sensitivity.

                                      Training and evaluation

                                      Inspection covered December 2022 to January 2023; operator analysis supported permanent implementation in 2023.

                                      Calibration and external performance

                                      ICIBI found that routing appeared generally accurate but identified limited first-line assurance. Complete confusion matrices and independent current validation were NOT_RETRIEVED.

                                      Base rate and error burden

                                      Base rate: NOT_RETRIEVED for each complexity class.

                                      False positives: Misrouting simple applications as complex can increase delay and scrutiny; measured current burden was NOT_RETRIEVED.

                                      False negatives: Misrouting complex applications as simple may create quality or fraud-control risk; measured current burden was NOT_RETRIEVED.

                                      Subgroups and missingness

                                      Equality reviews and nationality profiles were examined, but complete current subgroup error rates and applicant-level missingness effects were NOT_RETRIEVED.

                                      Human authority and automation bias

                                      Authority: Human decision makers decide applications on individual merit; CARS routes work and shapes workload and scrutiny.

                                      Risk: Routing can anchor complexity expectations before substantive review.

                                      Institutional action and feedback loop

                                      Action: Assignment to workflow and staff skill level; complex applications may receive additional checks and take longer.

                                      Loop: Fraud findings and operational outcomes can update profiles and bulk-data indicators.

                                      Notice, appeal, correction, deletion

                                      Applicants are not described as receiving individualized routing explanations or a distinct CARS appeal; visa decision remedies are separate. Routing-data correction and downstream repair were NOT_RETRIEVED.

                                      Affected-person or community evidence

                                      The ICIBI inspection explicitly did not examine the customer perspective.

                                      Independent outcome evidence

                                      Independent inspection plus operator response and permanent implementation statement.

                                      Unresolved question

                                      What are current error rates, delay effects, nationality impacts, and applicant remedy outcomes under the production version?

                                      Source supports: The operator’s system purpose, rules-based nature, and non-decision role.

                                      Source does not establish: Fairness, calibration, or no effect on applicants.

                                      Primary or authoritative records

                                      Correction trigger: A later ICIBI inspection, court record, or transparency update changes these findings.

                                      Reopening trigger: Impact assessment, audit, updated transparency record, or litigation.

                                      CDC FluSight — full outcome and evidence boundary
                                      Stable ID
                                      PRED-05-CDC-FLUSIGHT
                                      Jurisdiction
                                      United States
                                      Operator
                                      U.S. Centers for Disease Control and Prevention with forecasting partners
                                      Current status
                                      2025_26_SEASON_COMPLETE_PROGRAM_EXPECTED_TO_RESUME_FALL_2026
                                      Prior review
                                      2026-07-28
                                      Current review
                                      2026-07-28
                                      Source version
                                      WIP58_LONGITUDINAL_REFRESH_V2
                                      Currentness relation
                                      SUPPLEMENTS
                                      Validation status
                                      SEASONAL_EVALUATION_PRACTICE_DOCUMENTED_PERFORMANCE_VARIES_BY_SEASON_AND_HORIZON
                                      Specialist disposition
                                      NOT_COMPLETED

                                      Legal authority

                                      CDC public-health forecasting and planning function; forecasts are advisory planning information, not individual clinical decisions.

                                      Model or rules version

                                      Seasonal multi-model ensemble; component models and inclusion vary by season.

                                      Target and time horizon

                                      Target: Weekly influenza hospital admissions and related flu activity.

                                      Horizon: Near-term weekly forecasts, generally up to several weeks depending on target.

                                      Data and exclusions

                                      National and jurisdictional surveillance data; reporting lags and revisions affect forecasts. Individual clinical characteristics are outside the aggregate model.

                                      Training and evaluation

                                      Season-specific submissions and retrospective evaluation after observed outcomes become available.

                                      Calibration and external performance

                                      CDC publishes proper-scoring and interval evaluations. The 2024–25 ensemble failed to anticipate a rapid late-December/January increase; publication was delayed until intervals reliably captured observations. The 2023–24 ensemble outperformed most component models but also missed rapid changes.

                                      Base rate and error burden

                                      Base rate: Observed weekly admissions are directly reported; base-rate language is less applicable than time-varying incidence.

                                      False positives: Overforecasting can misallocate attention or resources; season-specific burden is reflected in scoring but downstream cost was NOT_RETRIEVED.

                                      False negatives: Underforecasting rapid growth can delay preparedness; CDC documented missed rapid changes.

                                      Subgroups and missingness

                                      Geographic performance varies; individual demographic subgroup predictions are not the system’s unit. Reporting completeness and revisions are material.

                                      Human authority and automation bias

                                      Authority: Public-health officials decide how to use forecasts and may delay publication when performance is inadequate.

                                      Risk: A visually precise ensemble can be overtrusted despite wide intervals and known difficulty with turning points.

                                      Institutional action and feedback loop

                                      Action: Planning, situational awareness, and messaging to health-care and public-health partners.

                                      Loop: Public-health behavior and reporting can alter subsequent observations; model performance does not directly create clinical labels.

                                      Notice, appeal, correction, deletion

                                      Public forecasts and evaluations are transparent; individual appeal and deletion are not applicable to aggregate outputs, but correction and version history remain important.

                                      Affected-person or community evidence

                                      No direct patient-level outcome evaluation attributable to FluSight was retrieved.

                                      Independent outcome evidence

                                      Operator-published season evaluations with explicit missed-event disclosure; independent decision-impact evaluation NOT_RETRIEVED.

                                      Unresolved question

                                      How often do forecasts change allocation or messaging, and what measurable benefit or harm follows?

                                      Source supports: Existence, planning purpose, and evaluation model.

                                      Source does not establish: Perfect accuracy or transferability to political or security prediction.

                                      Primary or authoritative records

                                      Correction trigger: CDC revises the season evaluation or an independent decision-impact study is published.

                                      Reopening trigger: Season-end evaluation, methodology change, or program status change.

                                      UNHCR forced-displacement forecasting — full outcome and evidence boundary
                                      Stable ID
                                      PRED-06-UNHCR-DISPLACEMENT
                                      Jurisdiction
                                      International
                                      Operator
                                      United Nations High Commissioner for Refugees and partners
                                      Current status
                                      2025_TECHNICAL_WORK_RECORDED
                                      Prior review
                                      2026-07-28
                                      Current review
                                      2026-07-28
                                      Source version
                                      WIP58_LONGITUDINAL_REFRESH_V2
                                      Currentness relation
                                      UNCHANGED
                                      Validation status
                                      INSTITUTIONAL_TECHNICAL_WORK_DOCUMENTED_OPERATIONAL_PERFORMANCE_REVIEW_NOT_CLOSED
                                      Specialist disposition
                                      NOT_COMPLETED

                                      Legal authority

                                      UNHCR humanitarian planning and protection mandate; predictive outputs support planning rather than adjudicating individual refugee status.

                                      Model or rules version

                                      Project Jetson historical models plus newer UNHCR predictive and nowcasting initiatives; no single current global production version was identified.

                                      Target and time horizon

                                      Target: Forced-displacement arrivals or departures and related humanitarian demand.

                                      Horizon: Varies by project; historical Jetson predicted regional displacement and newer systems include near-term planning and scenario horizons.

                                      Data and exclusions

                                      Historical displacement monitoring, operational data, crisis indicators, and other covariates. Data gaps are greatest where crisis conditions impair observation; individuals without recorded movement may be omitted.

                                      Training and evaluation

                                      Historical Somalia data covered seven years and 18 regions. A 2025 workshop describes later work nearing production, but a complete independent current evaluation was not retrieved.

                                      Calibration and external performance

                                      Operator historical material reported accurate predictions in 11 of 18 Somalia regions at one checkpoint; complete scoring, calibration, and independent replication were NOT_RETRIEVED.

                                      Base rate and error burden

                                      Base rate: Region- and period-specific displacement counts; no single base rate.

                                      False positives: Overprediction may misallocate scarce resources; quantified cost was NOT_RETRIEVED.

                                      False negatives: Underprediction may leave needs unmet; quantified harm was NOT_RETRIEVED.

                                      Subgroups and missingness

                                      Age, gender, disability, route, documentation, and minority-group performance were NOT_RETRIEVED; crisis-data missingness is a central limitation.

                                      Human authority and automation bias

                                      Authority: Humanitarian analysts and operational leaders retain planning and allocation authority.

                                      Risk: Forecast authority may displace local knowledge when quantitative outputs appear more precise than underlying data.

                                      Institutional action and feedback loop

                                      Action: Scenario planning, preparedness, supply-chain positioning, and operational prioritization.

                                      Loop: Resource placement, border policy, conflict, and reporting capacity can alter observed movement and future data.

                                      Notice, appeal, correction, deletion

                                      Aggregate forecasts do not create a direct individual appeal path; model governance, correction, and community participation remain relevant.

                                      Affected-person or community evidence

                                      Independent affected-community evidence linking forecast use to benefits, harms, consent, or exclusion was NOT_RETRIEVED.

                                      Independent outcome evidence

                                      Operator technical and policy records; independent operational-impact evaluation NOT_RETRIEVED.

                                      Unresolved question

                                      Did forecast use measurably improve assistance, reduce delay, or create geographic or demographic omission?

                                      Source supports: The institutional forecasting initiative and humanitarian-planning purpose.

                                      Source does not establish: Individual movement prediction or universal performance across crises.

                                      Primary or authoritative records

                                      Correction trigger: UNHCR or an independent evaluator publishes calibrated performance and decision-impact results.

                                      Reopening trigger: Operational evaluation, model-card publication, or revised technical guidance.

                                      European Commission JRC Dynamic Conflict Risk Model — full outcome and evidence boundary
                                      Stable ID
                                      PRED-07-JRC-CONFLICT-RISK
                                      Jurisdiction
                                      European Union / global analysis
                                      Operator
                                      European Commission Joint Research Centre
                                      Current status
                                      2025_TECHNICAL_REPORT_AS_PUBLISHED
                                      Prior review
                                      2026-07-28
                                      Current review
                                      2026-07-28
                                      Source version
                                      WIP58_LONGITUDINAL_REFRESH_V2
                                      Currentness relation
                                      UNCHANGED
                                      Validation status
                                      MODEL_AND_INTEGRATION_APPROACH_DOCUMENTED_OUT_OF_SAMPLE_AND_POLICY_EFFECT_REVIEW_OPEN
                                      Specialist disposition
                                      NOT_COMPLETED

                                      Legal authority

                                      European Commission Joint Research Centre analytical support and EEAS early-warning / horizon-scanning use.

                                      Model or rules version

                                      DCRM uses random forest and gradient-boosted tree models with annual retraining; exact current production parameters are not public.

                                      Target and time horizon

                                      Target: Subnational conflict onset or escalation risk.

                                      Horizon: Monthly updates with a six-month forecast horizon.

                                      Data and exclusions

                                      Conflict-event, structural, political, economic, geographic, and other predictors described in the technical report; internal operational results and some sensitive variables are not public.

                                      Training and evaluation

                                      Historical data from 1991 onward; annual retraining and tracking of prediction successes and failures are described.

                                      Calibration and external performance

                                      The report describes the modeling and validation approach but does not publish the internal model-result tables required for independent reproduction. Complete calibration metrics were NOT_RETRIEVED.

                                      Base rate and error burden

                                      Base rate: Conflict onset is a low-base-rate event at the administrative-area/month level; exact operating prevalence and threshold were NOT_RETRIEVED.

                                      False positives: False alarms may divert diplomatic or analytical attention; quantified burden and downstream actions were NOT_RETRIEVED.

                                      False negatives: Missed escalation may delay prevention; quantified burden was NOT_RETRIEVED.

                                      Subgroups and missingness

                                      Geographic coverage spans over 2,600 areas in 140 countries; performance by region, data quality, governance type, and conflict-reporting density was NOT_RETRIEVED.

                                      Human authority and automation bias

                                      Authority: Analysts and policymakers use the model within country prioritization and horizon scanning; it does not independently initiate intervention.

                                      Risk: Risk maps can anchor country prioritization and overshadow qualitative or local evidence.

                                      Institutional action and feedback loop

                                      Action: Analytical prioritization, horizon scanning, and potential early-warning follow-up.

                                      Loop: Policy attention and data collection may become concentrated in flagged places, changing future observation density.

                                      Notice, appeal, correction, deletion

                                      No individual score is published, but countries and communities may be affected by prioritization; public correction, challenge, and affected-community mechanisms were NOT_RETRIEVED.

                                      Affected-person or community evidence

                                      No community-linked outcome or participation evaluation was retrieved.

                                      Independent outcome evidence

                                      Primary JRC technical report; independent operational benefit or harm evaluation NOT_RETRIEVED.

                                      Unresolved question

                                      What are threshold-specific false alarm and miss rates, regional calibration, and measurable policy outcomes?

                                      Source supports: The model architecture and its role within a mixed-method early-warning process.

                                      Source does not establish: Certainty of conflict onset or strategic benefit from actions taken.

                                      Primary or authoritative records

                                      Correction trigger: JRC publishes corrected model details, external validation, or outcome evaluation.

                                      Reopening trigger: Performance evaluation, model revision, or public operational-use record.

                                      NINE HIGH-CONSEQUENCE GAP REVIEWS

                                      Show what the bounded search did not retrieve

                                      Open each record to see the exact status, source scope, known blind spots, prohibited inference, and the event that should reopen the search. An open gap is not converted into a negative factual claim.

                                      1. 01 FTC companion-chatbot inquiry outcomeONGOING_STUDY_RECONFIRMED_NO_PUBLIC_FINAL_FINDINGS_RETRIEVED_AS_OF_2026-07-28
                                        Latest registered follow-up search
                                        SEARCH-10-FTC-COMPANION-V2
                                        Relation
                                        SUPPLEMENTS
                                        Public wording changed
                                        NO

                                        Bounded finding

                                        April 2026 testimony confirms the study remained ongoing; no public final report, closure, settlement, or enforcement record was retrieved.

                                        What is supported

                                        Existence, scope, and continuing information-gathering status of the inquiry.

                                        What is not established

                                        Product-specific wrongdoing, prevalence, clinical causation, or final regulatory outcome.

                                        Known blind spots

                                        Nonpublic company responses and confidential FTC work are inaccessible.

                                        Dependent public claim: Public wording that the FTC is investigating companion safety.

                                        Prohibited inference: Do not describe an inquiry as a finding or enforcement action.

                                        Recorded searches and repositories

                                        Queries

                                        • site:ftc.gov 2026 testimony companion chatbots inquiry findings
                                        • site:ftc.gov AI chatbots acting as companions 6(b) report enforcement settlement closure

                                        Repositories and issuing bodies

                                        • Federal Trade Commission press releases, testimony, reports, cases, and legal library

                                        Retrieved authority records

                                        Correction trigger: FTC publishes a public report or corrects the status.

                                        Reopening trigger: Search again when a 6(b) report, case, settlement, closure notice, or testimony appears.

                                      2. 02 FCC Biden robocall payment, review, and follow-on dispositionDOJ_COLLECTION_REFERRAL_RETRIEVED_PAYMENT_OR_JUDICIAL_COLLECTION_NOT_RETRIEVED
                                        Latest registered follow-up search
                                        SEARCH-11-FCC-ROBOCALL-V2
                                        Relation
                                        SUPPLEMENTS
                                        Public wording changed
                                        YES

                                        Bounded finding

                                        The public wording now distinguishes the forfeiture order and DOJ referral from payment or collection.

                                        What is supported

                                        Final FCC forfeiture status and 2024 reporting context.

                                        What is not established

                                        Payment status after 2024, state-case outcomes, voter behavior, or election effect.

                                        Known blind spots

                                        Collection can proceed through channels not indexed in the bounded search.

                                        Dependent public claim: The incident produced a final federal administrative penalty.

                                        Prohibited inference: Do not call the penalty paid, collected, reversed, or judicially affirmed without a source.

                                        Recorded searches and repositories

                                        Queries

                                        • site:fcc.gov Steve Kramer robocall forfeiture DOJ collection payment
                                        • site:justice.gov Steve Kramer FCC forfeiture collection
                                        • Steve Kramer robocall forfeiture judicial review payment

                                        Repositories and issuing bodies

                                        • FCC enforcement, agency financial reports, DOJ releases, and federal court records

                                        Retrieved authority records

                                        Correction trigger: FCC, DOJ, or a court publishes a disposition.

                                        Reopening trigger: Repeat FCC, DOJ, PACER/public court, and state-attorney-general searches.

                                      3. 03 Meta ayahuasca downstream restorationPOST_RESTORATION_RANKING_STRIKE_RECOMMENDATION_AND_LOST_REACH_EVIDENCE_NOT_RETRIEVED
                                        Latest registered follow-up search
                                        SEARCH-12-META-AYAHUASCA-V2
                                        Relation
                                        UNCHANGED
                                        Public wording changed
                                        NO

                                        Bounded finding

                                        The decision still establishes reversal and restoration, not downstream ranking, recommendation, strike, or lost-reach repair.

                                        What is supported

                                        Authoritative appellate reversal and immediate content restoration.

                                        What is not established

                                        Complete downstream repair or audience recovery.

                                        Known blind spots

                                        Internal ranking and account-state systems are not publicly observable.

                                        Dependent public claim: A restored post is one remedy stage.

                                        Prohibited inference: Do not equate visible restoration with complete repair.

                                        Recorded searches and repositories

                                        Queries

                                        • site:oversightboard.com ayahuasca brew restoration ranking strike recommendation eligibility
                                        • site:transparency.meta.com ayahuasca oversight board implementation

                                        Repositories and issuing bodies

                                        • Meta Oversight Board decisions and implementation updates; Meta transparency reporting

                                        Retrieved authority records

                                        Correction trigger: Meta or the Oversight Board publishes system-level repair evidence.

                                        Reopening trigger: Seek implementation reports, account-state records, creator evidence, and independent audits.

                                      4. 04 Robodebt compensation, accountability, data repair, and repeat preventionADDITIONAL_COMPENSATION_AND_RESOLUTION_SCHEME_IMPLEMENTATION_RETRIEVED_COMPLETE_DOWNSTREAM_REPAIR_NOT_ESTABLISHED
                                        Prior search
                                        SEARCH-04-ROBODEBT
                                        Latest registered follow-up search
                                        SEARCH-13-ROBODEBT-V2
                                        Relation
                                        SUPPLEMENTS
                                        Public wording changed
                                        YES

                                        Bounded finding

                                        The current ledger records the approved $475 million settlement and open resolution scheme while preserving unresolved data propagation and repeated-error evidence.

                                        What is supported

                                        Multiple remedy and accountability stages.

                                        What is not established

                                        Complete compensation, complete data cleansing, prevention, or recovery from every financial and psychological consequence.

                                        Known blind spots

                                        Ongoing scheme administration and individual circumstances are not yet fully reported.

                                        Dependent public claim: Robodebt produced large-scale harm and multiple remedy mechanisms.

                                        Prohibited inference: Do not label settlement or payment as complete repair.

                                        Recorded searches and repositories

                                        Queries

                                        • site:servicesaustralia.gov.au robodebt settlement 2026 compensation
                                        • site:servicesaustralia.gov.au income apportionment resolution scheme 2026
                                        • site:nacc.gov.au Operation Myrtleford robodebt 2026

                                        Repositories and issuing bodies

                                        • Services Australia, Federal Court public information, National Anti-Corruption Commission, and official implementation records

                                        Retrieved authority records

                                        Correction trigger: Court, administrator, agency, or audit publishes outcome and downstream-correction evidence.

                                        Reopening trigger: Track payment completion, exclusions, complaints, automated-decision reforms, data correction, and repeat-error metrics.

                                      5. 05 C2PA adoption and credential survival in crisis chainsSTANDARD_CONFIRMS_METADATA_REMOVAL_AND_SOFT_BINDING_RECOVERY_REPRESENTATIVE_CRISIS_SURVIVAL_RATES_NOT_RETRIEVED
                                        Prior search
                                        SEARCH-05-C2PA
                                        Latest registered follow-up search
                                        SEARCH-14-C2PA-V2
                                        Relation
                                        SUPPLEMENTS
                                        Public wording changed
                                        YES

                                        Bounded finding

                                        Public wording now states that metadata can be removed and that durable recovery is not a universal survival guarantee.

                                        What is supported

                                        What C2PA claims, layered resilience measures, and independent security criticisms.

                                        What is not established

                                        Universal survival, truth of depicted events, or falsity of unsigned media.

                                        Known blind spots

                                        Platform implementations, re-encoding, screenshotting, and visible UI adoption vary and are incompletely public.

                                        Dependent public claim: Content credentials are one authentication layer.

                                        Prohibited inference: Do not treat credential presence as truth or absence as falsity.

                                        Recorded searches and repositories

                                        Queries

                                        • site:spec.c2pa.org 2.4 durable Content Credentials metadata removed watermark fingerprint
                                        • C2PA credential survival screenshot repost print scan crisis distribution study

                                        Repositories and issuing bodies

                                        • C2PA specification, explainer, implementation guidance, independent formal and field research

                                        Retrieved authority records

                                        Correction trigger: A peer-reviewed cross-platform field evaluation or corrected specification becomes available.

                                        Reopening trigger: Test signed assets across capture, editing, messaging, social platforms, screenshots, downloads, and archive workflows.

                                      6. 06 Foundation-model machine-unlearning exactnessAUDIT_RESEARCH_RETRIEVED_TRUE_FORGETTING_REMAINS_OPEN_PRODUCTION_EXACTNESS_NOT_RETRIEVED
                                        Latest registered follow-up search
                                        SEARCH-15-UNLEARNING-V2
                                        Relation
                                        SUPPLEMENTS
                                        Public wording changed
                                        YES

                                        Bounded finding

                                        The claim was narrowed to distinguish approximate behavior suppression from exact retraining-equivalent forgetting.

                                        What is supported

                                        The gap between source deletion, output suppression, approximate unlearning, and auditable forgetting.

                                        What is not established

                                        That no robust method exists, or that any named vendor failed a deletion duty.

                                        Known blind spots

                                        Production models and training data are proprietary; exact retraining references are rarely available.

                                        Dependent public claim: Claims about machine unlearning require method- and audit-specific evidence.

                                        Prohibited inference: Do not describe UI deletion, refusal, or model editing as exact unlearning.

                                        Recorded searches and repositories

                                        Queries

                                        • 2026 machine unlearning systematic audit truly forget foundation model
                                        • site:research.google 2026 auditing machine unlearning
                                        • production foundation model unlearning exact retraining equivalence audit

                                        Repositories and issuing bodies

                                        • Google Research, arXiv and conference proceedings, standards and deployed-model documentation

                                        Retrieved authority records

                                        Correction trigger: A production audit with retraining reference or bounded approximation guarantee is published.

                                        Reopening trigger: Search benchmark, vendor transparency, regulatory audit, and independent replication records.

                                      7. 07 Mental-state inference independent field validityNO_COMPLETE_INDEPENDENT_CONSTRUCT_SPECIFIC_FIELD_VALIDATION_RETRIEVED
                                        Latest registered follow-up search
                                        SEARCH-16-MENTAL-STATE-V2
                                        Relation
                                        UNCHANGED
                                        Public wording changed
                                        NO

                                        Bounded finding

                                        No public study was retrieved that closes ground truth, calibration, base rate, subgroup, generalization, action, and remedy for a named deployed construct.

                                        What is supported

                                        Proxy-detection and institutional-harm concerns; user perceptions in studied settings.

                                        What is not established

                                        Reliable knowledge of emotion, deception, loyalty, intent, or dangerousness.

                                        Known blind spots

                                        Proprietary systems, changing products, and weak ground truth limit evaluation.

                                        Dependent public claim: Observed signals are not the mind itself.

                                        Prohibited inference: Do not convert facial, vocal, gaze, neural, or interaction features into psychological fact without construct-valid evidence.

                                        Recorded searches and repositories

                                        Queries

                                        • independent field validation emotion recognition named construct calibration base rate subgroup 2026
                                        • workplace emotion AI independent field study construct validity outcome
                                        • border deception detection field validation base rate subgroup

                                        Repositories and issuing bodies

                                        • Peer-reviewed HCI and affective-science literature, regulator records, deployment audits, and operator documentation

                                        Retrieved authority records

                                        Correction trigger: A preregistered, independent, consequential field validation becomes available.

                                        Reopening trigger: Seek system-specific model cards, threshold metrics, subgroup performance, actions, appeals, and outcomes.

                                      8. 08 Ukraine deepfake recipient-level and strategic effectsARTIFACT_AND_RAPID_RESPONSE_RETRIEVED_RECIPIENT_BELIEF_FRONTLINE_BEHAVIOR_AND_STRATEGIC_EFFECT_NOT_RETRIEVED
                                        Latest registered follow-up search
                                        SEARCH-17-UKRAINE-DEEPFAKE-V2
                                        Relation
                                        UNCHANGED
                                        Public wording changed
                                        YES

                                        Bounded finding

                                        The public wording was narrowed to avoid treating ridicule or rapid rebuttal as proof of zero operational effect.

                                        What is supported

                                        Early incident and response stages plus online discourse.

                                        What is not established

                                        Uniform rejection, zero effect, or confirmed strategic success.

                                        Known blind spots

                                        Military communications and platform exposure data may be nonpublic or unavailable.

                                        Dependent public claim: The 2022 surrender deepfake was rapidly debunked and poorly received in observed online contexts.

                                        Prohibited inference: Do not convert ridicule or removal into proof of no recipient effect.

                                        Recorded searches and repositories

                                        Queries

                                        • Zelensky surrender deepfake recipient belief frontline behavior study
                                        • Ukraine Zelensky deepfake operational effect soldiers survey
                                        • Zelenskyy deepfake trust study recipient exposure

                                        Repositories and issuing bodies

                                        • Incident forensics, Ukrainian official records, independent journalism, peer-reviewed deepfake and trust research

                                        Retrieved authority records

                                        Correction trigger: Recipient surveys, military records, platform data, or attribution evidence becomes available.

                                        Reopening trigger: Search Ukrainian-language records, military archives, platform transparency, surveys, and oral-history evidence.

                                      9. 09 VIGINUM false positives, oversight outcomes, and affected-community experienceMISSION_CURRENTNESS_RETRIEVED_PUBLIC_FALSE_POSITIVE_REMEDY_AND_AFFECTED_COMMUNITY_REGISTER_NOT_RETRIEVED
                                        Prior search
                                        SEARCH-09-VIGINUM
                                        Latest registered follow-up search
                                        SEARCH-18-VIGINUM-V2
                                        Relation
                                        SUPPLEMENTS
                                        Public wording changed
                                        YES

                                        Bounded finding

                                        The page now records the 2026 remit change without upgrading operator reports into independent accuracy or remedy evidence.

                                        What is supported

                                        Institutional safeguards and self-reported operations.

                                        What is not established

                                        Error-free attribution, independent validation, or complete remedy.

                                        Known blind spots

                                        Some intelligence methods and corrections may be nonpublic.

                                        Dependent public claim: VIGINUM provides one behavior- and foreign-interference-focused democratic-defense model with published oversight features.

                                        Prohibited inference: Do not treat an institutional attribution as self-proving or infer zero false positives from absent public corrections.

                                        Recorded searches and repositories

                                        Queries

                                        • site:sgdsn.gouv.fr VIGINUM 2026 decree mission false positive correction appeal
                                        • site:sgdsn.gouv.fr VIGINUM ethics committee false positives remedy
                                        • VIGINUM affected community appeal correction false positive

                                        Repositories and issuing bodies

                                        • SGDSN/VIGINUM reports and mission pages, ethics committee opinions, legal records, oversight and affected-community publications

                                        Retrieved authority records

                                        Correction trigger: An independent audit, court case, complaint record, correction notice, or affected-community study becomes available.

                                        Reopening trigger: Repeat French-language searches across SGDSN, Legifrance, CNIL, courts, parliament, press, and civil society.

                                      CRISIS AUTHENTICATION

                                      Use overlapping controls—and state what each control cannot prove

                                      Authentication fails when one detector or one credential is treated as an oracle. Provenance, forensics, official confirmation, corrections, and public literacy answer different questions.

                                      Live confirmation

                                      Direct interactive confirmation can rapidly rebut a recording, but the channel itself must be authenticated and available.

                                      Official authentication channels

                                      Pre-established official channels reduce ambiguity; channel compromise or impersonation remains possible.

                                      Cross-channel redundancy

                                      Independent channels reduce single-point failure but do not guarantee audience exposure.

                                      Cryptographic signing and C2PA

                                      Can verify signed provenance claims; cannot prove the depicted event is true or make unsigned media false.

                                      Watermarking

                                      Can signal synthetic origin in supported systems; marks may be absent, stripped, or unavailable for open models.

                                      Forensic analysis

                                      Can identify artifacts or inconsistencies; detector confidence is not identity, sponsorship, or effect.

                                      Reverse search and source comparison

                                      Can reveal older or unaltered material; novel synthetic media may have no prior source.

                                      Public correction archives

                                      Preserve institutional memory; a correction record does not prove every audience received or believed it.

                                      Prebunking and public literacy

                                      Can prepare audiences for known tactics; must not become generalized distrust of all evidence.

                                      Rapid-response teams

                                      Can coordinate verification and communication; authority, oversight, privacy, and correction must be explicit.

                                      BEHAVIOR-BASED DEMOCRATIC DEFENSE

                                      Defend institutions without turning lawful belief into hostile terrain

                                      • Target concealed coordination, impersonation, technical tampering, foreign direction, and artificial amplification rather than lawful belief.
                                      • Separate platform behavior evidence from content truth judgments and political viewpoint.
                                      • Require attributable authority, narrow mandate, independent oversight, correction, and public reporting.
                                      • Protect lawful pseudonymity, journalism, research, activism, minority-language speech, and emergency coordination from false positives.
                                      • Do not treat institutional classification as self-proving; preserve alternative explanations and reopening triggers.

                                      ATTRIBUTABLE REVIEW STAGING

                                      Seven packets are assigned; zero reviews have started

                                      Prepared for review is not reviewed. A research report is not a specialist disposition. Submitted is not accepted. Validated is not endorsed. A reviewer assertion is not automatically the repository conclusion. A correction is not complete until affected public wording, durable memory, indexes, manifests, content tables, downstream derivatives, automated tests, and any required human retest are updated.

                                      Packets assigned
                                      7
                                      Reviews started
                                      0
                                      Real submissions
                                      0
                                      Schema-validated real submissions
                                      0
                                      Repository adjudications
                                      0
                                      Completed attributable dispositions
                                      0
                                      Public identity authorizations
                                      0
                                      Human testing
                                      PREPARED_NOT_EXECUTED
                                      Human participants
                                      0
                                      REV-CLINICAL-02PACKET_ASSIGNED_REVIEW_NOT_STARTED

                                      Clinical and psychological safety

                                      Review clinical framing, causal language, diagnostic boundaries, crisis and vulnerability wording, companion-system safety, reality-testing limits, and whether public material avoids diagnosing identifiable people.

                                      Packet assigned
                                      YES
                                      Review started
                                      NO
                                      Submissions
                                      0
                                      Schema validated
                                      0
                                      Repository adjudicated
                                      0
                                      Completed dispositions
                                      0
                                      Public identity authorized
                                      NO
                                      Human testing
                                      PREPARED_NOT_EXECUTED
                                      Scope, stable claims, and reopening trigger

                                      Stable claims: CLAIM-REAL-03-MENTAL-STATE, CLAIM-REAL-06-AFFECTED-EVIDENCE, CLAIM-REAL-07-COMPANION-SAFETY, CLAIM-REAL-08-INCIDENT-EFFECT, CLAIM-REAL-11-LIARS-DIVIDEND

                                      • Does the current wording for every assigned claim remain within the review scope of clinical and psychological safety?
                                      • Which claims should be accepted as worded, narrowed, corrected, deferred for more evidence, or marked outside reviewer scope?
                                      • Do the source relationships match the proposition being supported, including issuing-body, operator, independent, affected-person, and derivative evidence?

                                      Reopening trigger: An attributable review start, structurally valid submission, privacy-safe supersession, material new source, correction request, or authorized human-testing evidence is supplied.

                                      Prepared for review is not reviewed. Submitted is not accepted. Validated is not endorsed. A reviewer assertion is not automatically the repository conclusion. A correction is not complete until affected public wording, durable memory, indexes, manifests, content tables, and tests are updated.

                                      REV-SCIENCE-03PACKET_ASSIGNED_REVIEW_NOT_STARTED

                                      Scientific validity and measurement

                                      Review construct validity, calibration, generalization, base rates, subgroup burden, causal inference, effect attribution, measurement quality, auditability, and wording about what available evidence can and cannot establish.

                                      Packet assigned
                                      YES
                                      Review started
                                      NO
                                      Submissions
                                      0
                                      Schema validated
                                      0
                                      Repository adjudicated
                                      0
                                      Completed dispositions
                                      0
                                      Public identity authorized
                                      NO
                                      Human testing
                                      PREPARED_NOT_EXECUTED
                                      Scope, stable claims, and reopening trigger

                                      Stable claims: CLAIM-REAL-02-UNLEARNING, CLAIM-REAL-03-MENTAL-STATE, CLAIM-REAL-08-INCIDENT-EFFECT, CLAIM-REAL-09-SWARM-DETECTION, CLAIM-REAL-10-DEEPFAKE-RESPONSE, CLAIM-REAL-11-LIARS-DIVIDEND, CLAIM-REAL-13-PREDICTIVE

                                      • Does the current wording for every assigned claim remain within the review scope of scientific validity and measurement?
                                      • Which claims should be accepted as worded, narrowed, corrected, deferred for more evidence, or marked outside reviewer scope?
                                      • Do the source relationships match the proposition being supported, including issuing-body, operator, independent, affected-person, and derivative evidence?

                                      Reopening trigger: An attributable review start, structurally valid submission, privacy-safe supersession, material new source, correction request, or authorized human-testing evidence is supplied.

                                      Prepared for review is not reviewed. Submitted is not accepted. Validated is not endorsed. A reviewer assertion is not automatically the repository conclusion. A correction is not complete until affected public wording, durable memory, indexes, manifests, content tables, and tests are updated.

                                      REV-INTEL-04PACKET_ASSIGNED_REVIEW_NOT_STARTED

                                      Intelligence, attribution, and information-integrity analysis

                                      Review artifact status, coordination, actor identity, sponsorship, intent, origin-chain independence, authentication, correction timing, recipient evidence, and effect attribution without converting analytic confidence into fact.

                                      Packet assigned
                                      YES
                                      Review started
                                      NO
                                      Submissions
                                      0
                                      Schema validated
                                      0
                                      Repository adjudicated
                                      0
                                      Completed dispositions
                                      0
                                      Public identity authorized
                                      NO
                                      Human testing
                                      PREPARED_NOT_EXECUTED
                                      Scope, stable claims, and reopening trigger

                                      Stable claims: CLAIM-REAL-08-INCIDENT-EFFECT, CLAIM-REAL-09-SWARM-DETECTION, CLAIM-REAL-10-DEEPFAKE-RESPONSE, CLAIM-REAL-11-LIARS-DIVIDEND, CLAIM-REAL-12-DEMOCRATIC-DEFENSE

                                      • Does the current wording for every assigned claim remain within the review scope of intelligence, attribution, and information-integrity analysis?
                                      • Which claims should be accepted as worded, narrowed, corrected, deferred for more evidence, or marked outside reviewer scope?
                                      • Do the source relationships match the proposition being supported, including issuing-body, operator, independent, affected-person, and derivative evidence?

                                      Reopening trigger: An attributable review start, structurally valid submission, privacy-safe supersession, material new source, correction request, or authorized human-testing evidence is supplied.

                                      Prepared for review is not reviewed. Submitted is not accepted. Validated is not endorsed. A reviewer assertion is not automatically the repository conclusion. A correction is not complete until affected public wording, durable memory, indexes, manifests, content tables, and tests are updated.

                                      REV-PRIVACY-05PACKET_ASSIGNED_REVIEW_NOT_STARTED

                                      Privacy, data governance, and machine unlearning

                                      Review source-data and derived-data correction, retention, deletion, unlearning, downstream propagation, neural and behavioral data, private identity, lawful access, and technical claims about influence removal.

                                      Packet assigned
                                      YES
                                      Review started
                                      NO
                                      Submissions
                                      0
                                      Schema validated
                                      0
                                      Repository adjudicated
                                      0
                                      Completed dispositions
                                      0
                                      Public identity authorized
                                      NO
                                      Human testing
                                      PREPARED_NOT_EXECUTED
                                      Scope, stable claims, and reopening trigger

                                      Stable claims: CLAIM-REAL-01-REPAIR, CLAIM-REAL-02-UNLEARNING, CLAIM-REAL-03-MENTAL-STATE, CLAIM-REAL-04-LAW, CLAIM-REAL-05-VISIBILITY, CLAIM-REAL-13-PREDICTIVE

                                      • Does the current wording for every assigned claim remain within the review scope of privacy, data governance, and machine unlearning?
                                      • Which claims should be accepted as worded, narrowed, corrected, deferred for more evidence, or marked outside reviewer scope?
                                      • Do the source relationships match the proposition being supported, including issuing-body, operator, independent, affected-person, and derivative evidence?

                                      Reopening trigger: An attributable review start, structurally valid submission, privacy-safe supersession, material new source, correction request, or authorized human-testing evidence is supplied.

                                      Prepared for review is not reviewed. Submitted is not accepted. Validated is not endorsed. A reviewer assertion is not automatically the repository conclusion. A correction is not complete until affected public wording, durable memory, indexes, manifests, content tables, and tests are updated.

                                      REV-ACCESS-06PACKET_ASSIGNED_REVIEW_NOT_STARTED

                                      Accessibility, plain language, and inclusive design

                                      Review whether status, uncertainty, source scope, correction, evidence stages, and review boundaries are understandable and operable across assistive technologies, narrow screens, cognitive load, language, and literacy differences.

                                      Packet assigned
                                      YES
                                      Review started
                                      NO
                                      Submissions
                                      0
                                      Schema validated
                                      0
                                      Repository adjudicated
                                      0
                                      Completed dispositions
                                      0
                                      Public identity authorized
                                      NO
                                      Human testing
                                      PREPARED_NOT_EXECUTED
                                      Scope, stable claims, and reopening trigger

                                      Stable claims: CLAIM-REAL-01-REPAIR, CLAIM-REAL-04-LAW, CLAIM-REAL-05-VISIBILITY, CLAIM-REAL-06-AFFECTED-EVIDENCE, CLAIM-REAL-07-COMPANION-SAFETY, CLAIM-REAL-12-DEMOCRATIC-DEFENSE

                                      • Does the current wording for every assigned claim remain within the review scope of accessibility, plain language, and inclusive design?
                                      • Which claims should be accepted as worded, narrowed, corrected, deferred for more evidence, or marked outside reviewer scope?
                                      • Do the source relationships match the proposition being supported, including issuing-body, operator, independent, affected-person, and derivative evidence?

                                      Reopening trigger: An attributable review start, structurally valid submission, privacy-safe supersession, material new source, correction request, or authorized human-testing evidence is supplied.

                                      Prepared for review is not reviewed. Submitted is not accepted. Validated is not endorsed. A reviewer assertion is not automatically the repository conclusion. A correction is not complete until affected public wording, durable memory, indexes, manifests, content tables, and tests are updated.

                                      REV-AFFECTED-07PACKET_ASSIGNED_REVIEW_NOT_STARTED

                                      Affected-person and affected-community evidence ethics

                                      Review consent, selection, privacy, retaliation, withdrawal, representativeness, first-person and representative evidence, community framing, and whether lived experience is used without claiming prevalence or endorsement.

                                      Packet assigned
                                      YES
                                      Review started
                                      NO
                                      Submissions
                                      0
                                      Schema validated
                                      0
                                      Repository adjudicated
                                      0
                                      Completed dispositions
                                      0
                                      Public identity authorized
                                      NO
                                      Human testing
                                      PREPARED_NOT_EXECUTED
                                      Scope, stable claims, and reopening trigger

                                      Stable claims: CLAIM-REAL-01-REPAIR, CLAIM-REAL-03-MENTAL-STATE, CLAIM-REAL-05-VISIBILITY, CLAIM-REAL-06-AFFECTED-EVIDENCE, CLAIM-REAL-07-COMPANION-SAFETY, CLAIM-REAL-13-PREDICTIVE

                                      • Does the current wording for every assigned claim remain within the review scope of affected-person and affected-community evidence ethics?
                                      • Which claims should be accepted as worded, narrowed, corrected, deferred for more evidence, or marked outside reviewer scope?
                                      • Do the source relationships match the proposition being supported, including issuing-body, operator, independent, affected-person, and derivative evidence?

                                      Reopening trigger: An attributable review start, structurally valid submission, privacy-safe supersession, material new source, correction request, or authorized human-testing evidence is supplied.

                                      Prepared for review is not reviewed. Submitted is not accepted. Validated is not endorsed. A reviewer assertion is not automatically the repository conclusion. A correction is not complete until affected public wording, durable memory, indexes, manifests, content tables, and tests are updated.

                                      REAL-WORLD INTERPRETIVE

                                      ADVERSARIAL IMPORT HARDENING · PACKET V3

                                      Technical review intake is hardened; substantive review remains at zero

                                      WIP.62 accepts strict UTF-8 JSON or a constrained ZIP, rejects ambiguous or unsafe representations, quarantines private material, binds corrections and supersessions to exact prior hashes, and keeps reviewer assertions separate from repository adjudication.

                                      Adversarial fixture classes51

                                      Repository-authored non-evidence tests only.

                                      Lifecycle dry runs16

                                      No public claim was changed by a dry run.

                                      Real review submissions0

                                      Validation is not acceptance or endorsement.

                                      Accepted corrections0

                                      Correction closure requires synchronized public and memory updates.

                                      Review the implemented hardening controls
                                      • ZIP traversal, absolute path, backslash, duplicate, directory, symlink, non-regular, hidden-extra, casefold, and Unicode-normalization rejection
                                      • strict UTF-8, BOM, NUL, duplicate-key, malformed JSON, and non-finite number rejection
                                      • unknown-field and repository-assigned-state rejection
                                      • packet version, packet assignment, claim assignment, and duplicate-disposition checks
                                      • date ordering, future date, completion, and attributable-withdrawal checks
                                      • private attribution, public consent, affiliation, contact, conflict, sensitive data, and attachment quarantine
                                      • exact artifact, canonical JSON, and import-ID replay protection
                                      • hash-bound supersession, artifact-version increase, packet continuity, cycle, and correction-target checks
                                      • public-safe receipts that omit private values

                                      Human boundary: named reviewers, repository adjudications, accepted imports, completed specialist dispositions, participants, and sessions remain zero. Human testing remains PREPARED_NOT_EXECUTED.

                                      DETERMINISTIC PACKET EXPORTS

                                      Seven self-contained exports are prepared for attributable review

                                      Each export includes only track-relevant material, a versioned form and schema, privacy and correction instructions, a manifest, and checksums. Export preparation is an invitation—not review evidence.

                                      Protected review exports; no review start or disposition is inferred from their existence.
                                      PacketTrackClaimsMembersBytesSHA-256Status
                                      REV-LEGAL-01Legal and regulatory currentness51324,412e680168b7af5fbdbaea0dc239a5875bb73605732a74551033f0e7802c4025f0eDETERMINISTIC_V3_PACKET_PREPARED_REVIEW_NOT_STARTED
                                      This export is an invitation to review, not evidence that review occurred.
                                      REV-CLINICAL-02Clinical and psychological safety51321,449bb1e6164d724be6c9a73a02d05a00508ef3b32948df1f22645f9bd12cc5ec06aDETERMINISTIC_V3_PACKET_PREPARED_REVIEW_NOT_STARTED
                                      This export is an invitation to review, not evidence that review occurred.
                                      REV-SCIENCE-03Scientific validity and measurement71329,91655a18df16b89ed7351791862368fe483ce85f51e853b07a8db1b2dc5098de229DETERMINISTIC_V3_PACKET_PREPARED_REVIEW_NOT_STARTED
                                      This export is an invitation to review, not evidence that review occurred.
                                      REV-INTEL-04Intelligence, attribution, and information-integrity analysis51320,4192f9ca84597dd65fb128e9a69d152472a67e9a72e4fd58bf1cfa3768966119ab2DETERMINISTIC_V3_PACKET_PREPARED_REVIEW_NOT_STARTED
                                      This export is an invitation to review, not evidence that review occurred.
                                      REV-PRIVACY-05Privacy, data governance, and machine unlearning61326,148c8c499d665a4beaf24c5e9a8eb6c3cb3ec4ba2db67f405142c04bcc4242b7d90DETERMINISTIC_V3_PACKET_PREPARED_REVIEW_NOT_STARTED
                                      This export is an invitation to review, not evidence that review occurred.
                                      REV-ACCESS-06Accessibility, plain language, and inclusive design61320,63887a2bb3d8d03f890408d4ae3d6a856b77cf1b9d9040d777c4aad6b4519cd6854DETERMINISTIC_V3_PACKET_PREPARED_REVIEW_NOT_STARTED
                                      This export is an invitation to review, not evidence that review occurred.
                                      REV-AFFECTED-07Affected-person and affected-community evidence ethics61325,32709e524882771d1729237270d17518dbbbbf11b7e365a7f670b9b27504938f584DETERMINISTIC_V3_PACKET_PREPARED_REVIEW_NOT_STARTED
                                      This export is an invitation to review, not evidence that review occurred.

                                      PRIVACY, QUARANTINE & EXACT TEXT

                                      Identity is private by default; reviewer assertions remain separate from repository conclusions

                                      Private by default

                                      Attributable identity, affiliation, contact routes, attachments, and sensitive personal data stay in protected staging. Public identity requires granular affirmative consent and a separate repository decision.

                                      Exact assertion preserved

                                      The validator checks the SHA-256 of every reviewer assertion. A reviewer statement is retained as that reviewer’s evidence and is never silently rewritten into a repository conclusion.

                                      Validation is not acceptance

                                      Schema and semantic validation can route an artifact to pending adjudication or private quarantine. Neither route accepts a claim, publishes a correction, or creates certification.

                                      Limits remain explicit

                                      The repository does not promise legal privilege, absolute confidentiality, anonymity, deletion from backups, or control over copies already distributed downstream.

                                      Prepared for review is not reviewed. A research report is not a specialist disposition. Submitted is not accepted. Validated is not endorsed. A reviewer assertion is not automatically the repository conclusion. A correction is not complete until affected public wording, durable memory, indexes, manifests, content tables, downstream derivatives, automated tests, and any required human retest are updated.

                                      TEN-DIMENSION STATE MODEL

                                      18 required states keep review, adjudication, publication, and retesting separate

                                      No single badge can truthfully represent submission, schema, privacy, reviewer disposition, repository adjudication, public claim, correction, currentness, automated retest, and human retest at once.

                                      Submission Status

                                      NOT_STARTED · IN_REVIEW · SUBMITTED_PENDING_VALIDATION · WITHDRAW_PENDING_REVIEW · SUPERSEDED

                                      Schema Status

                                      NOT_STARTED · VALIDATED_PENDING_ADJUDICATION · REJECTED_INVALID

                                      Privacy Status

                                      NOT_STARTED · QUARANTINED_PRIVATE · PUBLIC_SAFE_DERIVATIVE_READY

                                      Reviewer Disposition

                                      NOT_STARTED · IN_REVIEW · ACCEPT_AS_WORDED · ACCEPT_WITH_NARROWING · REQUIRES_CORRECTION · REQUIRES_MORE_EVIDENCE · OUTSIDE_REVIEWER_SCOPE · WITHDRAW_PENDING_REVIEW

                                      Repository Adjudication

                                      NOT_STARTED · VALIDATED_PENDING_ADJUDICATION · REJECTED_OUT_OF_SCOPE · ACCEPT_AS_WORDED · ACCEPT_WITH_NARROWING · REQUIRES_CORRECTION · REQUIRES_MORE_EVIDENCE · CLOSED_WITHOUT_PUBLIC_CHANGE

                                      Public Claim State

                                      NOT_STARTED · PUBLIC_SAFE_DERIVATIVE_READY · PUBLIC_CORRECTION_PUBLISHED · CLOSED_WITHOUT_PUBLIC_CHANGE

                                      Correction State

                                      NOT_STARTED · REQUIRES_CORRECTION · PUBLIC_CORRECTION_PUBLISHED · CLOSED_WITHOUT_PUBLIC_CHANGE

                                      Currentness State

                                      NOT_STARTED · IN_REVIEW · RETEST_REQUIRED · CLOSED_WITHOUT_PUBLIC_CHANGE

                                      Automated Retest State

                                      NOT_STARTED · RETEST_REQUIRED · IN_REVIEW · CLOSED_WITHOUT_PUBLIC_CHANGE

                                      Human Retest State

                                      NOT_STARTED · RETEST_REQUIRED · PREPARED_NOT_EXECUTED · CLOSED_WITHOUT_PUBLIC_CHANGE

                                      Typed transitions with explicit origins, targets, effects, and reopening triggers.
                                      StepDimensionAllowed originsTargetPublic effectReopening trigger
                                      TRANS-01-RECEIVE
                                      RECEIVE                                      		submission_statusNOT_STARTED · IN_REVIEWSUBMITTED_PENDING_VALIDATIONNo public effect.Corrected or superseding artifact supplied.
                                      TRANS-02-SCHEMA_PASS
                                      SCHEMA_PASS                                      		schema_statusNOT_STARTEDVALIDATED_PENDING_ADJUDICATIONNo public effect.Schema or packet version changes.
                                      TRANS-03-SCHEMA_FAIL
                                      SCHEMA_FAIL                                      		schema_statusNOT_STARTEDREJECTED_INVALIDNo public effect.A corrected artifact is submitted.
                                      TRANS-04-QUARANTINE
                                      QUARANTINE                                      		privacy_statusNOT_STARTED · QUARANTINED_PRIVATEQUARANTINED_PRIVATENo public effect.Sensitive fields are removed or publication consent changes through a new superseding artifact.
                                      TRANS-05-REVIEW_START
                                      REVIEW_START                                      		reviewer_dispositionNOT_STARTEDIN_REVIEWPublic counter may show review started without identity.Reviewer withdraws, completes, or submits correction.
                                      TRANS-06-REVIEW_DISPOSITION
                                      REVIEW_DISPOSITION                                      		reviewer_dispositionIN_REVIEW · SUBMITTED_PENDING_VALIDATION · VALIDATED_PENDING_ADJUDICATION{REVIEWER_TERMINAL}No automatic public wording change.Repository adjudication or reviewer correction/supersession.
                                      TRANS-07-REPOSITORY_ADJUDICATE
                                      REPOSITORY_ADJUDICATE                                      		repository_adjudicationVALIDATED_PENDING_ADJUDICATION · IN_REVIEW{REPOSITORY_DECISION}Only accepted public-safe derivatives may proceed.New source, contradiction, correction, appeal, or currentness evidence.
                                      TRANS-08-PUBLIC_DERIVATIVE
                                      PUBLIC_DERIVATIVE                                      		privacy_statusQUARANTINED_PRIVATE · VALIDATED_PENDING_ADJUDICATION · ACCEPT_AS_WORDED · ACCEPT_WITH_NARROWING · REQUIRES_CORRECTIONPUBLIC_SAFE_DERIVATIVE_READYDerivative may be queued for publication after acceptance.Consent withdrawal, privacy issue, or new correction.
                                      TRANS-09-PUBLISH_CORRECTION
                                      PUBLISH_CORRECTION                                      		correction_stateREQUIRES_CORRECTION · PUBLIC_SAFE_DERIVATIVE_READYPUBLIC_CORRECTION_PUBLISHEDPublic wording changes with visible history.Downstream propagation failure, new evidence, or reviewer withdrawal.
                                      TRANS-10-RETEST
                                      RETEST                                      		automated_retest_stateNOT_STARTED · RETEST_REQUIRED · IN_REVIEWCLOSED_WITHOUT_PUBLIC_CHANGEMay close automated status only.Material code/content change or failed test.
                                      TRANS-11-HUMAN_RETEST_PENDING
                                      HUMAN_RETEST_PENDING                                      		human_retest_stateNOT_STARTED · RETEST_REQUIREDPREPARED_NOT_EXECUTEDPublicly states zero participants and no findings.Authorized attributable participant evidence supplied.
                                      TRANS-12-WITHDRAW
                                      WITHDRAW                                      		submission_statusIN_REVIEW · SUBMITTED_PENDING_VALIDATION · VALIDATED_PENDING_ADJUDICATIONWITHDRAW_PENDING_REVIEWPublic identity or assertion is not newly published.Withdrawal adjudication or superseding request.
                                      TRANS-13-SUPERSEDE
                                      SUPERSEDE                                      		submission_statusSUBMITTED_PENDING_VALIDATION · VALIDATED_PENDING_ADJUDICATION · IN_REVIEW · WITHDRAW_PENDING_REVIEW · REJECTED_INVALIDSUPERSEDEDNo automatic public effect.New superseding artifact or reopened decision.

                                      HUMAN EVALUATION READINESS

                                      Ten task scripts are executable but remain unexecuted

                                      Participants: 0. Recruitment, scheduling, compensation, recording, and human-certification authority have not been granted.

                                      HUMAN-AT-01-SCREEN-READER

                                      Screen-reader navigation

                                      Determine whether a screen-reader user can identify page purpose, navigate headings and landmarks, inspect review status, reach claim corrections, and understand tables without visual context.

                                      PREPARED_NOT_EXECUTED · 0 participants

                                      HUMAN-AT-02-KEYBOARD

                                      Keyboard-only navigation

                                      Determine whether every essential review, claim, source, correction, disclosure, and navigation control is reachable and operable without a pointing device.

                                      PREPARED_NOT_EXECUTED · 0 participants

                                      HUMAN-AT-03-REFLOW

                                      Magnification and reflow

                                      Determine whether evidence, tables, packet status, and forms remain readable at 200–400% zoom and at 320 CSS pixels without two-dimensional scrolling for ordinary text.

                                      PREPARED_NOT_EXECUTED · 0 participants

                                      HUMAN-AT-04-VOICE

                                      Voice-control navigation

                                      Determine whether visible labels, headings, links, and disclosure controls can be targeted predictably by speech-recognition software.

                                      PREPARED_NOT_EXECUTED · 0 participants

                                      HUMAN-AT-05-REDUCED-MOTION

                                      Reduced-motion and vestibular comfort

                                      Determine whether the pages remain comfortable with reduced-motion preferences and whether any transition, sticky element, scroll behavior, or animated emphasis causes discomfort.

                                      PREPARED_NOT_EXECUTED · 0 participants

                                      HUMAN-AT-06-COGNITIVE-LOAD

                                      Cognitive load and plain-language comprehension

                                      Determine whether status definitions, evidence boundaries, and correction steps can be understood without specialist vocabulary or excessive working-memory demand.

                                      PREPARED_NOT_EXECUTED · 0 participants

                                      HUMAN-AT-07-SOURCE-CORRECTION

                                      Source and correction discoverability

                                      Determine whether a reader can find the source scope, prior wording, current wording, contradiction record, correction trigger, and reopening trigger for a consequential claim.

                                      PREPARED_NOT_EXECUTED · 0 participants

                                      HUMAN-AT-08-MOBILE

                                      Mobile and narrow-screen use

                                      Determine whether review status, claim history, source relationships, and tables remain usable on touch devices and narrow screens.

                                      PREPARED_NOT_EXECUTED · 0 participants

                                      HUMAN-AT-09-MULTILINGUAL

                                      Multilingual and translation-access review

                                      Determine whether status labels, uncertainty, review boundaries, and correction instructions remain understandable when translated or read by multilingual users.

                                      PREPARED_NOT_EXECUTED · 0 participants

                                      HUMAN-AT-10-EVIDENCE-STATUS

                                      Evidence-status comprehension

                                      Determine whether a representative reader can distinguish prepared review from completed disposition and can separate evidence stages such as availability, exposure, belief, behavior, and effect.

                                      PREPARED_NOT_EXECUTED · 0 participants

                                      CLAIM COMPREHENSION

                                      Twenty-six distinctions that public wording must preserve

                                      No comprehension results are claimed. The eight WIP.62 additions test review assertion versus repository adjudication, validation versus acceptance, correction publication versus downstream repair, identity consent, packet assignment, consensus, currentness, and outcome evidence.

                                      Prepared comprehension checks; participant count is zero.
                                      IDDistinctionExpected conceptRevision trigger
                                      COMP-01-INQUIRY-FINDINGInquiry ≠ FindingAn inquiry asks questions or gathers information; a finding states a supported conclusion after an authorized process.A participant treats an inquiry, request, audit opening, or study as a final finding.
                                      COMP-02-ALLEGATION-ADJUDICATIONAllegation ≠ AdjudicationAn allegation is a claim that may be disputed; adjudication is a decision by an authorized tribunal or process.A participant cannot identify the procedural stage or assumes settlement equals merits judgment.
                                      COMP-03-ARTIFACT-ATTRIBUTIONArtifact existence ≠ Actor attributionShowing that an artifact exists does not identify who created, directed, funded, or intended it.A participant infers actor identity or sponsorship from content status alone.
                                      COMP-04-AVAILABILITY-EXPOSUREAvailability ≠ ExposureAvailable content can be accessed; exposure requires evidence that a person or audience actually encountered it.A participant equates hosting, indexing, or potential access with recipient exposure.
                                      COMP-05-EXPOSURE-BELIEFExposure ≠ BeliefSeeing content does not establish that the recipient believed it.A participant treats receipt, view, impression, or listening as acceptance.
                                      COMP-06-BELIEF-BEHAVIORBelief ≠ BehaviorA stated or measured belief does not automatically produce action.A participant infers behavior or strategic effect without measured action and counterfactual evidence.
                                      COMP-07-CAPABILITY-DEPLOYMENTTechnical capability ≠ DeploymentA system can demonstrate a capability in a lab without being deployed in a real institution or campaign.A participant cannot identify the deployment evidence or operating context.
                                      COMP-08-DEPLOYMENT-EFFECTDeployment ≠ Valid effectDeployment documents use; valid effect requires appropriate outcomes, measurement, and alternative explanations.A participant treats operator adoption, scale, or duration as causal validation.
                                      COMP-09-APPEAL-REPAIRAppeal or reversal ≠ Complete repairAn appeal can reverse an immediate decision while copied data, derived scores, lost opportunities, or repeated error remain.A participant cannot name at least one downstream repair dimension.
                                      COMP-10-DELETION-UNLEARNINGSource deletion ≠ Model unlearningDeleting source data does not automatically remove its learned influence from embeddings, derived features, caches, or model weights.A participant treats suppression, tombstoning, or output blocking as retraining-equivalent influence removal.
                                      COMP-11-SIGNAL-MINDSignal detection ≠ Knowledge of mental stateA sensor can detect movement, voice, gaze, physiology, or neural activity without validly knowing emotion, deception, loyalty, intent, or dangerousness.A participant collapses sensor or feature accuracy into construct validity.
                                      COMP-12-OPERATOR-VALIDATIONOperator description ≠ Independent validationAn operator can accurately describe its system or action while remaining non-independent for performance, fairness, or effect claims.A participant cannot distinguish institutional action evidence from independent evaluation.
                                      COMP-13-DECOMMISSION-REPAIRDecommissioning ≠ Deletion and repairStopping a system does not establish that scores, exports, copied labels, downstream records, or prior harms were corrected.A participant treats shutdown, replacement, or contract termination as complete deletion and remedy.
                                      COMP-14-PROVENANCE-TRUTHSigned provenance ≠ TruthA valid signature can authenticate a recorded origin and edit history; it does not prove the depicted event is true or appropriately interpreted.A participant treats cryptographic integrity as semantic truth or endorsement.
                                      COMP-15-MISSING-PROVENANCEMissing provenance ≠ FalsityUnsigned or stripped media may be authentic; absence of credentials is not proof of fabrication.A participant uses missing provenance as a falsity verdict.
                                      COMP-16-FIRST-PERSON-PREVALENCEFirst-person evidence ≠ PrevalenceA first-person account can establish an experienced event or harm but cannot by itself estimate how common it is.A participant generalizes illustrative testimony to a population rate.
                                      COMP-17-JURISDICTION-UNIVERSALOne jurisdiction’s law ≠ Universal ruleA law, decision, or guidance record applies within its jurisdiction and procedural status unless another authority extends it.A participant cannot identify jurisdiction, effective status, or whether a proposal is enacted.
                                      COMP-18-PREPARED-COMPLETEDPrepared review packet ≠ Completed specialist dispositionA prepared packet makes review executable; completed review requires an attributable reviewer, scope and conflict declaration, dated disposition, rationale, and accepted repository decision.A participant describes NOT_STARTED or IN_REVIEW as endorsement, certification, or accepted correction.
                                      COMP-19-REVIEWER-ASSERTION-REPOSITORY-ADJUDICATIONReviewer assertion ≠ Repository adjudicationA reviewer statement is preserved evidence from that reviewer; the repository separately checks sources, scope, privacy, and public wording before deciding.A participant collapses reviewer assertion into repository adjudication or cannot identify the additional evidence required.
                                      COMP-20-SCHEMA-VALID-FACTUALLY-ACCEPTEDSchema-valid ≠ Factually acceptedSchema-valid means required fields and types passed; factual acceptance requires source and scope adjudication.A participant collapses schema-valid into factually accepted or cannot identify the additional evidence required.
                                      COMP-21-CORRECTION-PUBLISHED-DOWNSTREAM-REPAIR-COMPLETECorrection published ≠ Downstream repair completePublishing corrected wording does not prove copied records, recommendations, income, rankings, or other downstream consequences were repaired.A participant collapses correction published into downstream repair complete or cannot identify the additional evidence required.
                                      COMP-22-PRIVATE-IDENTITY-RETAINED-PUBLIC-IDENTITY-AUTHORIZEDPrivate identity retained ≠ Public identity authorizedThe repository may retain attributable identity privately without permission to display it publicly.A participant collapses private identity retained into public identity authorized or cannot identify the additional evidence required.
                                      COMP-23-PACKET-ASSIGNED-REVIEW-STARTEDPacket assigned ≠ Review startedAssignment makes a packet available to a track; review starts only when attributable evidence records that work began.A participant collapses packet assigned into review started or cannot identify the additional evidence required.
                                      COMP-24-ONE-ACCEPTED-DISPOSITION-CONSENSUS-OR-CERTIFICATIONOne accepted disposition ≠ Consensus or certificationOne accepted review disposition is one attributable contribution, not field-wide consensus, institutional certification, or legal clearance.A participant collapses one accepted disposition into consensus or certification or cannot identify the additional evidence required.
                                      COMP-26-DOCUMENTED-ACTIVITY-INDEPENDENT-OUTCOME-EVIDENCEDocumented activity ≠ Independent outcome evidenceRecords of deployment, takedown, restoration, or agency action do not by themselves measure recipient belief, behavior, benefit, harm, or strategic effect.A participant collapses documented activity into independent outcome evidence or cannot identify the additional evidence required.

                                      REVIEW → ADJUDICATION → CORRECTION

                                      Sixteen attributable stages preserve disagreement, privacy, and supersession

                                      A reviewer recommendation is review-history evidence. It becomes public wording only after source verification, privacy review, repository adjudication, a public-safe derivative, all affected records, and required retesting.

                                      1. REVIEW-HARDENING-01Container admitted or rejectedJSON or constrained ZIP is classified before parsing.
                                      2. REVIEW-HARDENING-02Archive paths and resource limits checkedTraversal, collisions, duplicates, symlinks, hidden extras, size, and ratio limits are enforced.
                                      3. REVIEW-HARDENING-03Exact bytes preserved and hashedContainer and review payload hashes are recorded before interpretation.
                                      4. REVIEW-HARDENING-04Strict UTF-8 and duplicate-key parse completedBOM, mixed encoding, NUL, duplicate keys, and non-finite numbers are rejected.
                                      5. REVIEW-HARDENING-05Schema validation completedRequired fields and types are checked without implying factual acceptance.
                                      6. REVIEW-HARDENING-06Packet and claim assignment resolvedPacket version and assigned stable claim IDs are verified.
                                      7. REVIEW-HARDENING-07Identity and granular publication consent checkedPrivate attribution remains the default.
                                      8. REVIEW-HARDENING-08Sensitive data and attachments quarantinedContact, private identity, conflicts, and attachments remain protected.
                                      9. REVIEW-HARDENING-09Replay and duplicate defenses completedExact bytes, canonical JSON, and import IDs are checked.
                                      10. REVIEW-HARDENING-10Supersession and correction lineage checkedPrior ID, exact hash, packet, version, and cycle constraints are enforced.
                                      11. REVIEW-HARDENING-11Reviewer assertions preserved exactlySubmitted assertion bytes remain distinct from repository prose.
                                      12. REVIEW-HARDENING-12Candidate repository adjudication created separatelyThe repository may accept, narrow, correct, request evidence, reject, or close without adopting reviewer text.
                                      13. REVIEW-HARDENING-13Public-safe derivative prepared only after adjudicationPrivate values and attachments remain omitted.
                                      14. REVIEW-HARDENING-14Correction propagation dry-run executedPublic wording, tables, durable memory, indexes, manifests, and tests are treated as one closure set.
                                      15. REVIEW-HARDENING-15Automated retest recordedAutomated checks close only automated status.
                                      16. REVIEW-HARDENING-16Human retest state recordedHuman evaluation remains separately evidenced and PREPARED_NOT_EXECUTED unless real sessions occur.
                                      17. REVIEW-HARDENING-17Prior versions and disagreements retainedSuperseded, withdrawn, corrected, and disputed artifacts remain traceable.
                                      18. REVIEW-HARDENING-18Closure or reopening trigger recordedNo review lifecycle closes without an explicit reopening condition.
                                      OWNER-SUPPLIED RESEARCH INPUT — NOT SPECIALIST DISPOSITION

                                      SIX INDEPENDENT RESEARCH FOUNDATIONS

                                      Research input is preserved exactly and kept separate from review evidence

                                      The six WIP.63 reports deepen human evaluation, packet transport, correction propagation, privacy-preserving intake, adjudication governance, and claim currentness. Their source identity is verified; their citations and conclusions remain subject to claim-specific review.

                                      REVRES-01-HUMAN-EVALUATIONPREPARED_NOT_EXECUTED

                                      HUMAN EVALUATION AND ACCESSIBILITY

                                      Human Accessibility, Claim Comprehension, and Trauma-Informed Evaluation Protocol

                                      A mixed-method, trauma-informed evaluation design for testing whether readers preserve critical AI PSYOPS evidence distinctions across assistive-technology, technical-literacy, and affected-community contexts.

                                      What it adds
                                      Executable study architecture, cohort planning, task-based comprehension measures, recruitment and consent safeguards, stopping rules, data minimization, and retesting criteria.
                                      Evidence boundary
                                      The protocol is prepared but unexecuted. It contributes no participants, sessions, observations, accessibility certification, or empirical findings.
                                      Exact owner file
                                      AI Research Evaluation Protocol Design.md
                                      Bytes
                                      66,032
                                      SHA-256
                                      488164f01d1fc4a312579b6a80cc485f572bd87822c3c0538de268efa4b3c5b6
                                      REVRES-02-PACKET-INTEGRITYRESEARCH_FOUNDATION_SOURCE_IDENTITY_VERIFIED

                                      PACKET INTEGRITY AND TRANSPORT

                                      Offline Review-Packet Integrity, Provenance, and Replay-Resistant Transport

                                      A defensive comparison of integrity, authenticity, confidentiality, deterministic serialization, archive safety, replay resistance, signatures, and offline verification practices.

                                      What it adds
                                      A minimum integrity profile, a higher-assurance signed profile, a packet threat model, failure taxonomy, privacy checklist, and non-developer verification workflow.
                                      Evidence boundary
                                      Hash matching establishes integrity, not truth or endorsement. A signature authenticates a key relationship, not a reviewer’s substantive accuracy or independence.
                                      Exact owner file
                                      Offline Packet Integrity Transport.md
                                      Bytes
                                      47,187
                                      SHA-256
                                      2d6f9d1f01fde5c8ebe6d907033365b81c1ecd80a8a5e0c9cc2071d6b0a94c97
                                      REVRES-03-CORRECTION-PROPAGATIONRESEARCH_FOUNDATION_SOURCE_IDENTITY_VERIFIED

                                      CORRECTION CLOSURE AND SYSTEM REPAIR

                                      Correction Propagation, Retraction, Supersession, and Downstream Repair

                                      A systems-level account of why a correction notice is not complete repair across webpages, PDFs, feeds, search indexes, structured data, citations, archives, retrieval systems, embeddings, and model-training pipelines.

                                      What it adds
                                      A correction-status taxonomy, propagation map, transactional closure checklist, downstream repair matrix, discoverability framework, and reopening rules.
                                      Evidence boundary
                                      A posted correction does not prove downstream receipt, index refresh, belief revision, derived-data repair, model unlearning, or prevention of repeated error.
                                      Exact owner file
                                      Correction Propagation And System Repair.md
                                      Bytes
                                      49,623
                                      SHA-256
                                      532de28769e79352e609ea02f1657c4299da1ae0ff7d2eeeda511400afcd285c
                                      REVRES-04-PRIVACY-INTAKERESEARCH_FOUNDATION_SOURCE_IDENTITY_VERIFIED

                                      PRIVACY, CONSENT, AND PROTECTED IDENTITY

                                      Privacy-Preserving Reviewer Intake, Consent, and Protected Identity Governance

                                      A comparative review of confidential intake systems and layered controls for qualifications, identity, attribution consent, conflicts, sensitive evidence, metadata leakage, retention, and withdrawal.

                                      What it adds
                                      A data-minimization matrix, layered consent model, sensitive-field classification, retention schedule, identity-verification options, and public/private receipt design.
                                      Evidence boundary
                                      No system can promise absolute anonymity, privilege, deletion from all backups, or control over copies already distributed. Cryptographic approaches require specialist implementation and threat-model review.
                                      Exact owner file
                                      Privacy-Preserving Reviewer Intake Systems.md
                                      Bytes
                                      52,627
                                      SHA-256
                                      05003880bb7b0e3a071deecbbfac0ea3850b69067f7dc8391d7cbf5cb2ce4a10
                                      REVRES-05-ADJUDICATION-GOVERNANCERESEARCH_FOUNDATION_SOURCE_IDENTITY_VERIFIED

                                      REVIEW AND ADJUDICATION GOVERNANCE

                                      Attributable Specialist Review and Claim-Adjudication Governance

                                      A cross-domain governance model for qualifications, assignment, conflicts, dispositions, minority reports, withdrawal, supersession, and separation between reviewer advice and institutional conclusions.

                                      What it adds
                                      A complete review lifecycle, adjudication taxonomy, separation-of-authority matrix, conflict model, privacy framework, and disagreement-preserving decision process.
                                      Evidence boundary
                                      The report describes governance patterns; it is not itself an attributable specialist review of PsychologicalWar.org and creates no accepted disposition.
                                      Exact owner file
                                      Review Adjudication Governance Research.md
                                      Bytes
                                      38,916
                                      SHA-256
                                      167eccaf82b02f66bdc32cad5987f97443bc829d6afe155323f70d3a5b894ff8
                                      REVRES-06-CLAIM-LIFECYCLERESEARCH_FOUNDATION_SOURCE_IDENTITY_VERIFIED

                                      CURRENTNESS AND CLAIM LIFECYCLE

                                      Dynamic Currentness, Legal Status, Scientific Consensus, and Claim Supersession

                                      A bitemporal model for tracking laws, standards, litigation, enforcement, policies, scientific consensus, publication status, amendments, withdrawal, and supersession without collapsing distinct procedural states.

                                      What it adds
                                      A cross-domain lifecycle taxonomy, source-authority hierarchy, currentness risk model, scheduled and event-driven review policy, contradiction model, and public wording patterns.
                                      Evidence boundary
                                      A proposal is not law, adoption is not application, publication is not implementation, a settlement is not a merits judgment, and one jurisdiction’s rule is not universal doctrine.
                                      Exact owner file
                                      Public Claim Lifecycle Research.md
                                      Bytes
                                      44,347
                                      SHA-256
                                      dd8ec075ac357b1b90c0a3ab3a03ebbf6202d9199a67641cc823e20654840dd4

                                      CLAIM-LEVEL ADJUDICATION WORKBENCH

                                      Reviewer advice, technical validation, repository judgment, and owner authority remain separate

                                      Prepared for review is not reviewed. Submitted is not accepted. Validated is not endorsed. A reviewer assertion is not automatically the repository conclusion. A visible correction notice is not complete repair. Closure requires current public wording, durable records, indexes, manifests, structured data, automated retests, and any authorized human retest.

                                      Reviewer

                                      May: Provide attributable evidence, reasoning, uncertainty, scope limits, and a recommendation.

                                      May not: Assign repository acceptance, publication, correction closure, or certification.

                                      Validator

                                      May: Check schema, packet integrity, lineage, hashes, privacy routing, and state consistency.

                                      May not: Endorse factual accuracy or resolve scientific, legal, clinical, or editorial disputes.

                                      Repository adjudicator

                                      May: Compare evidence, preserve disagreement, decide public-safe wording, and assign correction requirements.

                                      May not: Rewrite reviewer assertions or conceal conflicting evidence.

                                      Publisher/owner

                                      May: Authorize release within documented scope.

                                      May not: Convert owner approval into specialist certification, legal clearance, or human accessibility certification.

                                      Dry-run claim-diff ledger

                                      Every row is marked TEST_FIXTURE_NOT_REVIEW_EVIDENCE. These examples test wording boundaries and do not modify a live claim.

                                      Non-evidence before/after wording fixtures.
                                      IDChangePrior wordingBounded wordingRetest state
                                      DIFF-63-01
                                      TEST_FIXTURE_NOT_REVIEW_EVIDENCE                                      		SOURCE_NARROWINGThe report proves a complete review.The report proposes a review framework; no review was executed.CLOSED_WITHOUT_PUBLIC_CHANGE
                                      PREPARED_NOT_EXECUTED
                                      DIFF-63-02
                                      TEST_FIXTURE_NOT_REVIEW_EVIDENCE                                      		LEGAL_STATUS_NARROWINGThe instrument is universally binding.The instrument’s legal force depends on instrument type, jurisdiction, entry into force, applicability, and implementation.CLOSED_WITHOUT_PUBLIC_CHANGE
                                      PREPARED_NOT_EXECUTED
                                      DIFF-63-03
                                      TEST_FIXTURE_NOT_REVIEW_EVIDENCE                                      		CORRECTION_CLOSURE_NARROWINGThe webpage correction completed repair.The webpage correction is one propagation step; downstream records and required retests remain separately tracked.CLOSED_WITHOUT_PUBLIC_CHANGE
                                      PREPARED_NOT_EXECUTED
                                      DIFF-63-04
                                      TEST_FIXTURE_NOT_REVIEW_EVIDENCE                                      		SIGNATURE_BOUNDARYA signed packet is true and endorsed.A valid signature links the packet to a key; substantive truth, independence, and endorsement require separate review.CLOSED_WITHOUT_PUBLIC_CHANGE
                                      PREPARED_NOT_EXECUTED
                                      DIFF-63-05
                                      TEST_FIXTURE_NOT_REVIEW_EVIDENCE                                      		PRIVACY_BOUNDARYAnonymous intake guarantees anonymity.The intake minimizes and protects identity data but cannot promise absolute anonymity or control of external copies.CLOSED_WITHOUT_PUBLIC_CHANGE
                                      PREPARED_NOT_EXECUTED
                                      DIFF-63-06
                                      TEST_FIXTURE_NOT_REVIEW_EVIDENCE                                      		HUMAN_EVIDENCE_BOUNDARYThe accessibility protocol passed.The accessibility protocol is PREPARED_NOT_EXECUTED and contains no participant findings.CLOSED_WITHOUT_PUBLIC_CHANGE
                                      PREPARED_NOT_EXECUTED

                                      TRANSACTIONAL CORRECTION CLOSURE

                                      A correction closes only when every required propagation and retest target closes

                                      8 open or reopened fixtures and 1 complete-closure fixture demonstrate fail-closed behavior. All nine are infrastructure tests, not live corrections.

                                      WIP.63 correction-closure fixtures marked TEST_FIXTURE_NOT_REVIEW_EVIDENCE.
                                      FixtureCaseCompletedMissing or failedStateClosure allowed
                                      CLOSE-63-01PARTIAL_PROPAGATIONpublic wording updateddurable memory · manifest · automated retestOPENNO
                                      CLOSE-63-02STALE_PUBLIC_WORDINGdurable correction recordpublic wordingOPENNO
                                      CLOSE-63-03STALE_DURABLE_MEMORYpublic wording updateddurable memory · semantic pointersOPENNO
                                      CLOSE-63-04STALE_MANIFESTpublic and durable wording updatedcontent manifest · release manifestOPENNO
                                      CLOSE-63-05FAILED_AUTOMATED_RETESTall propagation targets updatedautomated retest passOPENNO
                                      CLOSE-63-06HUMAN_RETEST_REQUIREDautomated checks passedauthorized human retestOPEN_PREPARED_NOT_EXECUTEDNO
                                      CLOSE-63-07REOPENED_CORRECTIONprior closure recordednew evidence adjudicationREOPENEDNO
                                      CLOSE-63-08SUPERSEDED_CORRECTIONnew correction acceptedsupersession graph closureOPENNO
                                      CLOSE-63-09COMPLETE_CLOSUREall required targets, receipts, automated tests, and authorized human tests completeCLOSEDYES

                                      OFFLINE PACKET INTEGRITY & REPLAY RESISTANCE

                                      Integrity, authenticity, confidentiality, truth, and endorsement are different properties

                                      The current deterministic exports implement a minimum integrity profile. A higher-assurance signed profile is documented as research and is not represented as deployed.

                                      TRANSPORT-63-MINIMUMIMPLEMENTED_FOR_DETERMINISTIC_PACKET_EXPORTS

                                      Minimum integrity profile

                                      • deterministic ordering and timestamps
                                      • UTF-8 canonical data representation
                                      • member whitelist
                                      • manifest-bound SHA-256
                                      • safe paths
                                      • no symlinks
                                      • duplicate and Unicode-collision rejection
                                      • resource limits
                                      • offline checksum verification

                                      Establishes: The received bytes match the hashed packet structure under the stated canonicalization rules.

                                      Does not establish: Sender authenticity, confidentiality, reviewer qualifications, independence, factual accuracy, or endorsement.

                                      TRANSPORT-63-SIGNEDRESEARCHED_NOT_CLAIMED_AS_DEPLOYED

                                      Higher-assurance signed profile

                                      • all minimum-profile controls
                                      • detached signature over canonical manifest
                                      • documented trust anchor
                                      • key rotation and revocation record
                                      • signature timestamp policy
                                      • offline public-key verification
                                      • separate encryption decision

                                      Establishes: A configured verification key signed the canonical manifest at a recorded stage, subject to trust-anchor and revocation assumptions.

                                      Does not establish: Substantive truth, reviewer independence, legal privilege, confidentiality without encryption, or institutional endorsement.

                                      TRANSPORT-65-TEST-SIGNEDTEST_PROFILE_IMPLEMENTED_NOT_PRODUCTION_TRUST

                                      Test-only higher-assurance Ed25519 packet profile

                                      • restricted deterministic manifest canonicalization
                                      • payload SHA-256
                                      • detached Ed25519 signature
                                      • test public-key fingerprint
                                      • key-state registry
                                      • offline verification
                                      • semantic validation after cryptographic validation

                                      Establishes: For a PASS fixture, the configured test key signed the exact canonical manifest and the payload hashes match.

                                      Does not establish: Truth, quality, endorsement, reviewer qualification, independence, confidentiality, encryption, publication approval, or production trust.

                                      DYNAMIC CURRENTNESS · VERSION 6

                                      Eighteen lifecycle states prevent procedural steps from becoming stronger public claims

                                      A proposal is not law; enactment is not applicability; publication is not implementation; an inquiry is not a finding; a settlement is not automatically a merits judgment; one jurisdiction is not universal doctrine.

                                      PROPOSED

                                      Formally introduced but not binding or adopted.

                                      ADOPTED

                                      Accepted by an authority but not necessarily effective or applicable.

                                      ENACTED

                                      Completed formal enactment; provisions may have later effective dates.

                                      IN_FORCE

                                      Legally or institutionally active from an identified date.

                                      APPLICABLE

                                      In force and factually within the relevant actor, territory, subject, and exemption scope.

                                      IMPLEMENTED

                                      Operational measures have been deployed; publication alone is insufficient.

                                      ENFORCED

                                      A competent authority has applied or penalized non-compliance.

                                      CHALLENGED

                                      A formal dispute exists; the underlying rule may remain active unless stayed.

                                      STAYED

                                      Application is temporarily suspended by competent order.

                                      SETTLED

                                      Parties resolved a dispute without automatically establishing merits precedent.

                                      APPEALED

                                      A lower decision is under review; effect depends on jurisdiction and stay status.

                                      FINALLY_ADJUDICATED

                                      The competent process has reached a final disposition within its jurisdiction.

                                      AMENDED

                                      Specific provisions or claims changed; provision-level tracking is required.

                                      REPEALED

                                      Future force is ended by competent authority; historical application may remain relevant.

                                      WITHDRAWN

                                      The issuing body removed the item from active reliance or publication.

                                      SUPERSEDED

                                      A newer version replaces the older item for current use.

                                      ARCHIVED

                                      Retained for history and audit but explicitly non-current.

                                      NOT_RETRIEVED_OR_UNRESOLVED

                                      The bounded search cannot establish current status; definitive claims are prohibited.

                                      Seven selective official currentness records
                                      Bounded official-source status checks; every stronger inference remains prohibited.
                                      AuthorityRecordObserved statusSupportsDoes not support
                                      RFC EditorRFC 8493: The BagIt File Packaging Format (V1.0)
                                      CURRENT63-RFC8493                                      		PUBLISHED_INFORMATIONAL_RFC
                                      Official RFC Editor publication remains available.                                      		BagIt is an established file-packaging specification.That BagIt alone provides signatures, confidentiality, review acceptance, or factual endorsement.
                                      RFC EditorRFC 8785: JSON Canonicalization Scheme (JCS)
                                      CURRENT63-RFC8785                                      		PUBLISHED_INFORMATIONAL_RFC
                                      Official RFC Editor publication remains available.                                      		JCS defines a canonical JSON representation suitable for deterministic hashing or signing workflows.That canonicalization validates truth or reviewer authority.
                                      NISTNIST SP 800-161 Rev. 1
                                      CURRENT63-NIST-161R1                                      		FINAL_PUBLICATION
                                      Official NIST final-publication page remains available.                                      		Cybersecurity supply-chain risk management includes provenance and integrity controls.A claim that one NIST publication certifies this repository or packet profile.
                                      W3CVerifiable Credential Data Integrity 1.0
                                      CURRENT63-W3C-DATA-INTEGRITY                                      		W3C_RECOMMENDATION
                                      Official W3C Recommendation dated 15 May 2025 is available.                                      		W3C publishes a standards-track data-integrity mechanism for verifiable credentials.That PsychologicalWar.org uses or is certified under this specification.
                                      UNESCORecommendation on the Ethics of Neurotechnology
                                      CURRENT63-UNESCO-NEUROTECH                                      		ADOPTED_NON_BINDING_RECOMMENDATION
                                      UNESCO records adoption on 11 November 2025.                                      		A global normative recommendation on neurotechnology ethics exists.That the recommendation is directly binding law or uniformly implemented in every jurisdiction.
                                      Council of EuropeFramework Convention on Artificial Intelligence and Human Rights, Democracy and the Rule of Law
                                      CURRENT63-COE-AI-CONVENTION                                      		OPENED_FOR_SIGNATURE_STATUS_REQUIRES_TREATY_CHART
                                      Official Council of Europe materials record opening for signature on 5 September 2024.                                      		The treaty text and signature process exist.Universal entry into force, ratification, private-sector applicability, or domestic implementation without jurisdiction-specific verification.
                                      European UnionEU ratification of Council of Europe AI Convention
                                      CURRENT63-EU-COE-RATIFICATION                                      		OFFICIAL_EU_RATIFICATION_RECORD
                                      Official EU material records ratification on 15 May 2026.                                      		The EU completed its identified ratification act.That every state, private actor, or factual situation is immediately and uniformly governed without applicability and implementation analysis.
                                      REAL-WORLD INTERPRETIVE

                                      THIRTEEN LIVE CLAIM OPERATIONS

                                      Every consequential claim now has a reviewable operational record

                                      Stable claim IDs, exact current wording hashes, source tiers, source independence, conflicts, consent, valid time, transaction time, propagation targets, retests, closure, and reopening are recorded without inventing a reviewer or disposition.

                                      Live claim operations. Reviewer assertion remains NOT_SUPPLIED for every record.
                                      ClaimCurrent decisionSource stateReview stateRetest and reopening
                                      CLAIM-REAL-01-REPAIR
                                      Reversal is not complete repair
                                      c053dbcc70fd5629d622f12034ce0b5080ec0f44153def96c4b647b1cd88a1bc                                      		RETAIN_CURRENT_BOUNDED_WORDING_PENDING_REAL_REVIEW
                                      No genuine attributable review artifact was supplied. Current source-bounded wording and explicit limits remain in force.                                      		INDEPENDENT_PLATFORM_APPEAL_DECISION · OFFICIAL_ANTI_CORRUPTION_INVESTIGATION_REPORT · OFFICIAL_INQUIRY_REPORT · OFFICIAL_PARLIAMENTARY_ANSWER · OFFICIAL_REMEDY_PROGRAM_NOTICE · OFFICIAL_SETTLEMENT_IMPLEMENTATION_NOTICE
                                      INDEPENDENT_FOR_LIMITED_PROPOSITION · NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD · PARTIAL_OR_UNRESOLVED                                      		Assertion: NOT_SUPPLIED
                                      COI: NOT_APPLICABLE_NO_REVIEWER
                                      Correction: NOT ACCEPTED                                      		REQUIRED_ON_ACCEPTED_CHANGE
                                      New audit, compensation, propagation, or repeat-error data becomes available.
                                      CLAIM-REAL-02-UNLEARNING
                                      Deletion and output blocking are not model unlearning
                                      68276f0748259f644274db399fd2f54e3fdea5b981e7c18081c971807332e9b0                                      		RETAIN_CURRENT_BOUNDED_WORDING_PENDING_REAL_REVIEW
                                      No genuine attributable review artifact was supplied. Current source-bounded wording and explicit limits remain in force.                                      		ACADEMIC_PREPRINT_SYSTEMATIC_AUDIT · OFFICIAL_TECHNICAL_DEFINITION · PEER_REVIEWED_MACHINE_UNLEARNING_EVALUATION · PRIMARY_GOVERNMENT_TECHNICAL_GUIDANCE · PRIMARY_RESEARCH_ANNOUNCEMENT · PRIMARY_RESEARCH_PREPRINT
                                      INDEPENDENT_FOR_LIMITED_PROPOSITION · NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD · PARTIAL_OR_UNRESOLVED                                      		Assertion: NOT_SUPPLIED
                                      COI: NOT_APPLICABLE_NO_REVIEWER
                                      Correction: NOT ACCEPTED                                      		REQUIRED_ON_ACCEPTED_CHANGE
                                      Validated exact-unlearning benchmarks for large deployed models become available.
                                      CLAIM-REAL-03-MENTAL-STATE
                                      Observed movement is not direct knowledge of emotion
                                      8ab000762a883875380f4d8d7d02dd9693161799449b51a5be32c3724613d4c5                                      		RETAIN_CURRENT_BOUNDED_WORDING_PENDING_REAL_REVIEW
                                      No genuine attributable review artifact was supplied. Current source-bounded wording and explicit limits remain in force.                                      		ENACTED_LEGISLATION · PEER_REVIEWED_SCIENTIFIC_REVIEW · PEER_REVIEWED_SYSTEMS_AND_WORKPLACE_ANALYSIS · PEER_REVIEWED_USER_STUDY
                                      INDEPENDENT_FOR_LIMITED_PROPOSITION · PARTIAL_OR_UNRESOLVED                                      		Assertion: NOT_SUPPLIED
                                      COI: NOT_APPLICABLE_NO_REVIEWER
                                      Correction: NOT ACCEPTED                                      		REQUIRED_ON_ACCEPTED_CHANGE
                                      New scientific consensus or regulatory guidance materially changes the boundary.
                                      CLAIM-REAL-04-LAW
                                      Cognitive-liberty law is jurisdiction-specific and uneven
                                      5dd0fc5f0305048f8f836ef0a56162a6e0831bf4ebbef1065e429c55166cc3ff                                      		RETAIN_CURRENT_BOUNDED_WORDING_PENDING_REAL_REVIEW
                                      No genuine attributable review artifact was supplied. Current source-bounded wording and explicit limits remain in force.                                      		ENACTED_LEGISLATION · ENACTED_STATE_LEGISLATION
                                      PARTIAL_OR_UNRESOLVED                                      		Assertion: NOT_SUPPLIED
                                      COI: NOT_APPLICABLE_NO_REVIEWER
                                      Correction: NOT ACCEPTED                                      		REQUIRED_ON_ACCEPTED_CHANGE
                                      Implementing rules, enforcement, or appellate decisions become available.
                                      CLAIM-REAL-05-VISIBILITY
                                      Visibility actions and remedy outcomes are separate
                                      cb2fa8aaa0207a90deffb4195a0d359429f28c4c4ab1e2d263c987dd7f4d1a62                                      		RETAIN_CURRENT_BOUNDED_WORDING_PENDING_REAL_REVIEW
                                      No genuine attributable review artifact was supplied. Current source-bounded wording and explicit limits remain in force.                                      		INDEPENDENT_PLATFORM_APPEAL_DECISION
                                      PARTIAL_OR_UNRESOLVED                                      		Assertion: NOT_SUPPLIED
                                      COI: NOT_APPLICABLE_NO_REVIEWER
                                      Correction: NOT ACCEPTED                                      		REQUIRED_ON_ACCEPTED_CHANGE
                                      Audit data reveals downstream ranking, strike, monetization, or profile changes.
                                      CLAIM-REAL-06-AFFECTED-EVIDENCE
                                      Lived experience and institutional evidence require explicit scope
                                      d55f61ab535195e1e73271068dc0c61a0ccccd02000ea63ee3054b4edf71e2e8                                      		RETAIN_CURRENT_BOUNDED_WORDING_PENDING_REAL_REVIEW
                                      No genuine attributable review artifact was supplied. Current source-bounded wording and explicit limits remain in force.                                      		OFFICIAL_INQUIRY_REPORT · OFFICIAL_REMEDY_PROGRAM_NOTICE · OFFICIAL_SETTLEMENT_IMPLEMENTATION_NOTICE · PEER_REVIEWED_USER_STUDY
                                      INDEPENDENT_FOR_LIMITED_PROPOSITION · NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD                                      		Assertion: NOT_SUPPLIED
                                      COI: NOT_APPLICABLE_NO_REVIEWER
                                      Correction: NOT ACCEPTED                                      		REQUIRED_ON_ACCEPTED_CHANGE
                                      Authorized community research, representative surveys, or new official records become available.
                                      CLAIM-REAL-07-COMPANION-SAFETY
                                      Regulatory inquiry is not clinical causality
                                      1b3e56c698a1acdaa35491f0f6ef35f715176c8bee093dea594ca69b6f5f0209                                      		RETAIN_CURRENT_BOUNDED_WORDING_PENDING_REAL_REVIEW
                                      No genuine attributable review artifact was supplied. Current source-bounded wording and explicit limits remain in force.                                      		OFFICIAL_CONGRESSIONAL_TESTIMONY · OFFICIAL_CURRENTNESS_STATEMENT · OFFICIAL_REGULATORY_INQUIRY
                                      NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD                                      		Assertion: NOT_SUPPLIED
                                      COI: NOT_APPLICABLE_NO_REVIEWER
                                      Correction: NOT ACCEPTED                                      		REQUIRED_ON_ACCEPTED_CHANGE
                                      Peer-reviewed longitudinal evidence, final enforcement, or adjudicated product records become available.
                                      CLAIM-REAL-08-INCIDENT-EFFECT
                                      An incident record does not automatically prove behavioral effect
                                      cacb10c7d46a7b62e06cb46edf7c9dde9873ed6857acad9a22f96a055dd7253e                                      		RETAIN_CURRENT_BOUNDED_WORDING_PENDING_REAL_REVIEW
                                      No genuine attributable review artifact was supplied. Current source-bounded wording and explicit limits remain in force.                                      		INDEPENDENT_OPEN_SOURCE_INCIDENT_ANALYSIS · OFFICIAL_AGENCY_FINANCIAL_REPORT · OFFICIAL_ENFORCEMENT_REPORT · OFFICIAL_FINAL_ENFORCEMENT_NOTICE · PLATFORM_THREAT_INTELLIGENCE_REPORT
                                      INDEPENDENT_FOR_LIMITED_PROPOSITION · NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD · PARTIAL_OR_UNRESOLVED                                      		Assertion: NOT_SUPPLIED
                                      COI: NOT_APPLICABLE_NO_REVIEWER
                                      Correction: NOT ACCEPTED                                      		REQUIRED_ON_ACCEPTED_CHANGE
                                      Survey, telecom, or election research measures exposure, belief, or behavior.
                                      CLAIM-REAL-09-SWARM-DETECTION
                                      Behavioral coordination evidence is stronger than a synthetic-content score alone
                                      de23adfd1d91b4cfdee1ae7c5a218092e6fc9f43ad5c2224076dbe99a14514b6                                      		RETAIN_CURRENT_BOUNDED_WORDING_PENDING_REAL_REVIEW
                                      No genuine attributable review artifact was supplied. Current source-bounded wording and explicit limits remain in force.                                      		OFFICIAL_ACTIVITY_AND_GOVERNANCE_REPORT · OFFICIAL_ELECTION_PERIOD_REPORT · PLATFORM_THREAT_INTELLIGENCE_REPORT · PLATFORM_THREAT_REPORT
                                      NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD · PARTIAL_OR_UNRESOLVED                                      		Assertion: NOT_SUPPLIED
                                      COI: NOT_APPLICABLE_NO_REVIEWER
                                      Correction: NOT ACCEPTED                                      		REQUIRED_ON_ACCEPTED_CHANGE
                                      Independent forensic, legal, or financial evidence establishes actor or sponsor identity.
                                      CLAIM-REAL-10-DEEPFAKE-RESPONSE
                                      Official debunking closes some stages, not all effects
                                      569cca2f52b0142fd96622e11cb125bb20feb5271bd46b80b1861108bf1bc6c2                                      		RETAIN_CURRENT_BOUNDED_WORDING_PENDING_REAL_REVIEW
                                      No genuine attributable review artifact was supplied. Current source-bounded wording and explicit limits remain in force.                                      		INDEPENDENT_OPEN_SOURCE_INCIDENT_ANALYSIS · OFFICIAL_CYBERSECURITY_GUIDANCE · OFFICIAL_INCIDENT_REPORT · PEER_REVIEWED_SOCIAL_MEDIA_DISCOURSE_STUDY · TECHNICAL_STANDARD
                                      INDEPENDENT_FOR_LIMITED_PROPOSITION · NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD · PARTIAL_OR_UNRESOLVED                                      		Assertion: NOT_SUPPLIED
                                      COI: NOT_APPLICABLE_NO_REVIEWER
                                      Correction: NOT ACCEPTED                                      		REQUIRED_ON_ACCEPTED_CHANGE
                                      Platform telemetry, intelligence release, or recipient research measures exposure and effect.
                                      CLAIM-REAL-11-LIARS-DIVIDEND
                                      Provenance is not truth, and the liar’s dividend is context-dependent
                                      2d422bad94e7ddf4c4604b8bf5bb0090dd446ce95510b76d0c5e0b1221f2a022                                      		RETAIN_CURRENT_BOUNDED_WORDING_PENDING_REAL_REVIEW
                                      No genuine attributable review artifact was supplied. Current source-bounded wording and explicit limits remain in force.                                      		INDEPENDENT_FORMAL_SECURITY_ANALYSIS · OFFICIAL_CYBERSECURITY_GUIDANCE · PEER_REVIEWED_EXPERIMENTAL_RESEARCH · PEER_REVIEWED_SOCIAL_MEDIA_DISCOURSE_STUDY · PRIMARY_TECHNICAL_STANDARD · PRIMARY_TECHNICAL_STANDARD_EXPLAINER · TECHNICAL_STANDARD
                                      INDEPENDENT_FOR_LIMITED_PROPOSITION · NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD · PARTIAL_OR_UNRESOLVED                                      		Assertion: NOT_SUPPLIED
                                      COI: NOT_APPLICABLE_NO_REVIEWER
                                      Correction: NOT ACCEPTED                                      		REQUIRED_ON_ACCEPTED_CHANGE
                                      Field evidence on credential adoption, attack resistance, or liar’s-dividend effects becomes available.
                                      CLAIM-REAL-12-DEMOCRATIC-DEFENSE
                                      Democratic defense should target behavior and foreign interference, not lawful belief
                                      cccec9f9d5e2a028d68e8ef326be53b858974845dbb3469ee942475bbb4bcaeb                                      		RETAIN_CURRENT_BOUNDED_WORDING_PENDING_REAL_REVIEW
                                      No genuine attributable review artifact was supplied. Current source-bounded wording and explicit limits remain in force.                                      		INDEPENDENT_OR_SEPARATE_OVERSIGHT_OPINION · OFFICIAL_ACTIVITY_AND_GOVERNANCE_REPORT · OFFICIAL_ELECTION_PERIOD_REPORT · OFFICIAL_MISSION_AND_LEGAL_CURRENTNESS_PAGE · OFFICIAL_THREAT_ASSESSMENT · PRIMARY_REGULATORY_OPINION
                                      INDEPENDENT_FOR_LIMITED_PROPOSITION · NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD                                      		Assertion: NOT_SUPPLIED
                                      COI: NOT_APPLICABLE_NO_REVIEWER
                                      Correction: NOT ACCEPTED                                      		REQUIRED_ON_ACCEPTED_CHANGE
                                      Independent audits, litigation, parliamentary review, or affected-community evidence becomes available.
                                      CLAIM-REAL-13-PREDICTIVE
                                      Aggregate forecasting and person-level risk scoring are different system classes
                                      29a9e76a2e19f7873f7c4d4928217bf67f9537af912014a2e359fe4066e863ba                                      		RETAIN_CURRENT_BOUNDED_WORDING_PENDING_REAL_REVIEW
                                      No genuine attributable review artifact was supplied. Current source-bounded wording and explicit limits remain in force.                                      		INDEPENDENT_GOVERNMENT_INSPECTION · JUDICIAL_DECISION · OFFICIAL_ALGORITHMIC_TRANSPARENCY_RECORD · OFFICIAL_CURRENT_PROGRAM_PLAN · OFFICIAL_CURRENT_PROGRAM_RESPONSE · OFFICIAL_CURRENT_SEASON_STATUS · OFFICIAL_OPERATOR_EVALUATION · OFFICIAL_OPERATOR_RESPONSE · OFFICIAL_OVERSIGHT_ADVISORY · OFFICIAL_PARLIAMENTARY_ANSWER · OFFICIAL_POLICY_AND_PROGRAM_DESCRIPTION · OFFICIAL_PUBLIC_HEALTH_FORECASTING_PROGRAM · OFFICIAL_RESEARCH_REPORT · OFFICIAL_RESEARCH_SUMMARY · OFFICIAL_SYSTEM_DESCRIPTION · OFFICIAL_TECHNICAL_DESCRIPTION · OFFICIAL_TECHNICAL_WORKSHOP_RECORD · OPERATOR_COMMISSIONED_IMPACT_EVALUATION · PEER_REVIEWED_OR_ACADEMIC_EMPIRICAL_STUDY
                                      INDEPENDENT_FOR_LIMITED_PROPOSITION · NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD · PARTIAL_OR_UNRESOLVED                                      		Assertion: NOT_SUPPLIED
                                      COI: NOT_APPLICABLE_NO_REVIEWER
                                      Correction: NOT ACCEPTED                                      		REQUIRED_ON_ACCEPTED_CHANGE
                                      Independent evaluation, litigation, audit, or affected-person evidence becomes available.

                                      Authority boundary: Prepared for review is not reviewed. Submitted is not accepted. Validated is not endorsed. A reviewer assertion is not automatically the repository conclusion. A visible correction notice is not complete repair. Closure requires current public wording, durable records, indexes, manifests, structured data, automated retests, and any authorized human retest.

                                      PUBLIC CORRECTION DISCOVERABILITY

                                      Current wording, change history, limits, and reopening remain findable

                                      A correction route is visible for every claim. Repository status does not imply that external caches, citations, datasets, embeddings, or models have updated.

                                      CLAIM-REAL-01-REPAIRCURRENT_BOUNDED_WORDING_REVIEW_OPEN

                                      Reversal is not complete repair

                                      A successful appeal, reversal, settlement, or restored account can stop an immediate error without repairing every copied record, derived score, ranking signal, vendor system, financial loss, reputational consequence, or psychological harm. Federal Court approval on 23 June 2026 of an additional Robodebt settlement documents compensation and remedy stages, not proof of complete downstream repair.

                                      Last change
                                      CORR-REAL-01-REPAIR · 2026-07-28
                                      Specialist review
                                      NOT_COMPLETED
                                      Downstream state
                                      CURRENT_WIP64_TABLE_AND_DURABLE_REGISTER

                                      Submit evidence about this claim

                                      This status describes repository wording and propagation. It does not establish that every external cache, citation, dataset, embedding, or model has updated.

                                      CLAIM-REAL-02-UNLEARNINGCURRENT_BOUNDED_WORDING_REVIEW_OPEN

                                      Deletion and output blocking are not model unlearning

                                      Source deletion, suppression, output blocking, model editing, retraining, exact unlearning, and approximate unlearning are distinct interventions. Current standards and public research do not establish that a named production foundation model has exactly removed a requested influence or reached independently audited retraining-equivalent behavior across hidden and observable pathways.

                                      Last change
                                      CORR-REAL-02-UNLEARNING · 2026-07-28
                                      Specialist review
                                      NOT_COMPLETED
                                      Downstream state
                                      CURRENT_WIP64_TABLE_AND_DURABLE_REGISTER

                                      Submit evidence about this claim

                                      This status describes repository wording and propagation. It does not establish that every external cache, citation, dataset, embedding, or model has updated.

                                      CLAIM-REAL-03-MENTAL-STATECURRENT_BOUNDED_WORDING_REVIEW_OPEN

                                      Observed movement is not direct knowledge of emotion

                                      A system may detect a facial movement, vocal feature, gaze pattern, physiological signal, or interaction trace without reliably establishing the person’s emotion, deception, loyalty, vulnerability, intent, or dangerousness. Independent construct-specific field validity remains incomplete.

                                      Last change
                                      CORR-REAL-03-MENTAL-STATE · 2026-07-28
                                      Specialist review
                                      NOT_COMPLETED
                                      Downstream state
                                      CURRENT_WIP64_TABLE_AND_DURABLE_REGISTER

                                      Submit evidence about this claim

                                      This status describes repository wording and propagation. It does not establish that every external cache, citation, dataset, embedding, or model has updated.

                                      CLAIM-REAL-04-LAWCURRENT_BOUNDED_WORDING_REVIEW_OPEN

                                      Cognitive-liberty law is jurisdiction-specific and uneven

                                      Cognitive-liberty, neural-data, biometric, automated-decision, and AI-manipulation rules remain jurisdiction-specific, staged, and uneven. A proposal, guidance document, settlement, trial judgment, or one jurisdiction’s enacted rule is not a universal code.

                                      Last change
                                      CORR-REAL-04-LAW · 2026-07-28
                                      Specialist review
                                      NOT_COMPLETED
                                      Downstream state
                                      CURRENT_WIP64_TABLE_AND_DURABLE_REGISTER

                                      Submit evidence about this claim

                                      This status describes repository wording and propagation. It does not establish that every external cache, citation, dataset, embedding, or model has updated.

                                      CLAIM-REAL-05-VISIBILITYCURRENT_BOUNDED_WORDING_REVIEW_OPEN

                                      Visibility actions and remedy outcomes are separate

                                      Removal, restriction, search exclusion, recommendation exclusion, downranking, labeling, monetization change, account penalty, personalized refusal, synthesized reframing, and persistent-profile change are different actions. A visible restoration does not by itself show that ranking, recommendation, strikes, income, or downstream records were repaired.

                                      Last change
                                      CORR-REAL-05-VISIBILITY · 2026-07-28
                                      Specialist review
                                      NOT_COMPLETED
                                      Downstream state
                                      CURRENT_WIP64_TABLE_AND_DURABLE_REGISTER

                                      Submit evidence about this claim

                                      This status describes repository wording and propagation. It does not establish that every external cache, citation, dataset, embedding, or model has updated.

                                      CLAIM-REAL-06-AFFECTED-EVIDENCECURRENT_BOUNDED_WORDING_REVIEW_OPEN

                                      Lived experience and institutional evidence require explicit scope

                                      First-person, representative, institutional, aggregated, anonymized, and derivative evidence can document different parts of an AI-governance harm. None should be silently converted into prevalence, universal causation, representativeness, or affected-community endorsement.

                                      Last change
                                      CORR-REAL-06-AFFECTED-EVIDENCE · 2026-07-28
                                      Specialist review
                                      NOT_COMPLETED
                                      Downstream state
                                      CURRENT_WIP64_TABLE_AND_DURABLE_REGISTER

                                      Submit evidence about this claim

                                      This status describes repository wording and propagation. It does not establish that every external cache, citation, dataset, embedding, or model has updated.

                                      CLAIM-REAL-07-COMPANION-SAFETYCURRENT_BOUNDED_WORDING_REVIEW_OPEN

                                      Regulatory inquiry is not clinical causality

                                      The FTC’s companion-chatbot work remained publicly described as an ongoing Section 6(b) study in April 2026. An inquiry documents regulatory concern and information gathering; it is not a completed finding of wrongdoing, clinical causation, or prevalence.

                                      Last change
                                      CORR-REAL-07-COMPANION-SAFETY · 2026-07-28
                                      Specialist review
                                      NOT_COMPLETED
                                      Downstream state
                                      CURRENT_WIP64_TABLE_AND_DURABLE_REGISTER

                                      Submit evidence about this claim

                                      This status describes repository wording and propagation. It does not establish that every external cache, citation, dataset, embedding, or model has updated.

                                      CLAIM-REAL-08-INCIDENT-EFFECTCURRENT_BOUNDED_WORDING_REVIEW_OPEN

                                      An incident record does not automatically prove behavioral effect

                                      A verified artifact, coordinated distribution, operator record, or enforcement action can establish an incident without proving recipient exposure, belief, intention, behavior change, operational outcome, or strategic effect.

                                      Last change
                                      CORR-REAL-08-INCIDENT-EFFECT · 2026-07-28
                                      Specialist review
                                      NOT_COMPLETED
                                      Downstream state
                                      CURRENT_WIP64_TABLE_AND_DURABLE_REGISTER

                                      Submit evidence about this claim

                                      This status describes repository wording and propagation. It does not establish that every external cache, citation, dataset, embedding, or model has updated.

                                      CLAIM-REAL-09-SWARM-DETECTIONCURRENT_BOUNDED_WORDING_REVIEW_OPEN

                                      Behavioral coordination evidence is stronger than a synthetic-content score alone

                                      Behavioral and infrastructure evidence can establish coordination more strongly than an AI-content score alone. Coordination still does not automatically establish actor identity, sponsorship, intent, audience effect, or the synthetic origin of every asset.

                                      Last change
                                      CORR-REAL-09-SWARM-DETECTION · 2026-07-28
                                      Specialist review
                                      NOT_COMPLETED
                                      Downstream state
                                      CURRENT_WIP64_TABLE_AND_DURABLE_REGISTER

                                      Submit evidence about this claim

                                      This status describes repository wording and propagation. It does not establish that every external cache, citation, dataset, embedding, or model has updated.

                                      CLAIM-REAL-10-DEEPFAKE-RESPONSECURRENT_BOUNDED_WORDING_REVIEW_OPEN

                                      Official debunking closes some stages, not all effects

                                      The Zelenskyy surrender deepfake and rapid rebuttal establish the artifact, hacked distribution, and authentication response. Public sources reviewed here do not establish representative recipient belief, frontline behavior change, operational outcome, or strategic effect.

                                      Last change
                                      CORR-REAL-10-DEEPFAKE-RESPONSE · 2026-07-28
                                      Specialist review
                                      NOT_COMPLETED
                                      Downstream state
                                      CURRENT_WIP64_TABLE_AND_DURABLE_REGISTER

                                      Submit evidence about this claim

                                      This status describes repository wording and propagation. It does not establish that every external cache, citation, dataset, embedding, or model has updated.

                                      CLAIM-REAL-11-LIARS-DIVIDENDCURRENT_BOUNDED_WORDING_REVIEW_OPEN

                                      Provenance is not truth, and the liar’s dividend is context-dependent

                                      C2PA 2.4 supports signed provenance claims, durable credentials, soft bindings, manifest recovery, and trust-list governance. Credentials can still be absent, stripped, untrusted, or semantically misleading; they do not prove the depicted event is true, and their absence does not make media false. Liar’s-dividend effects remain context-dependent.

                                      Last change
                                      CORR-REAL-11-LIARS-DIVIDEND · 2026-07-28
                                      Specialist review
                                      NOT_COMPLETED
                                      Downstream state
                                      CURRENT_WIP64_TABLE_AND_DURABLE_REGISTER

                                      Submit evidence about this claim

                                      This status describes repository wording and propagation. It does not establish that every external cache, citation, dataset, embedding, or model has updated.

                                      CLAIM-REAL-12-DEMOCRATIC-DEFENSECURRENT_BOUNDED_WORDING_REVIEW_OPEN

                                      Democratic defense should target behavior and foreign interference, not lawful belief

                                      Democratic cognitive defense should target concealed coordination, foreign direction, impersonation, technical tampering, artificial amplification, fraud, and platform abuse rather than lawful belief. Current VIGINUM and CNIL records document institutional activity, oversight, retention limits, and indirect access, rectification, and erasure routes—not a public register proving false-positive frequency or completed remedy outcomes.

                                      Last change
                                      CORR-REAL-12-DEMOCRATIC-DEFENSE · 2026-07-28
                                      Specialist review
                                      NOT_COMPLETED
                                      Downstream state
                                      CURRENT_WIP64_TABLE_AND_DURABLE_REGISTER

                                      Submit evidence about this claim

                                      This status describes repository wording and propagation. It does not establish that every external cache, citation, dataset, embedding, or model has updated.

                                      CLAIM-REAL-13-PREDICTIVECURRENT_BOUNDED_WORDING_REVIEW_OPEN

                                      Aggregate forecasting and person-level risk scoring are different system classes

                                      Aggregate forecasting, case routing, family or person-level risk scoring, and institutional triage are different system classes. Current AFST, CARS, and FluSight records establish continued program activity, not universal validity, fairness, causal benefit, notice, appeal, or complete repair.

                                      Last change
                                      CORR-REAL-13-PREDICTIVE · 2026-07-28
                                      Specialist review
                                      NOT_COMPLETED
                                      Downstream state
                                      CURRENT_WIP64_TABLE_AND_DURABLE_REGISTER

                                      Submit evidence about this claim

                                      This status describes repository wording and propagation. It does not establish that every external cache, citation, dataset, embedding, or model has updated.

                                      FAIL-CLOSED CORRECTION TRANSACTIONS · V2

                                      Closure is a repository-wide transaction, not a local edit

                                      15 fixture cases remain open and 1 complete fixture closes. All 16 records are marked TEST_FIXTURE_NOT_REVIEW_EVIDENCE.

                                      Sixteen-target correction transaction dry runs.
                                      FixtureCaseFailed or missingHuman retestClosure
                                      CLOSURE-64-01MISSING_PUBLIC_WORDINGpublic_wordingNOT REQUIREDBLOCKED
                                      CLOSURE-64-02STALE_PUBLIC_PROVENANCEpublic_provenanceNOT REQUIREDBLOCKED
                                      CLOSURE-64-03STALE_CONTENT_TABLEcontent_tableNOT REQUIREDBLOCKED
                                      CLOSURE-64-04STALE_DURABLE_REGISTERdurable_claim_registerNOT REQUIREDBLOCKED
                                      CLOSURE-64-05STALE_BRIEFSdocs_report_briefsNOT REQUIREDBLOCKED
                                      CLOSURE-64-06STALE_POINTER_LEDGERsemantic_pointer_ledgerNOT REQUIREDBLOCKED
                                      CLOSURE-64-07STALE_PROTECTED_INDEXprotected_domain_indexNOT REQUIREDBLOCKED
                                      CLOSURE-64-08STALE_REPORT_MANIFESTreport_manifestNOT REQUIREDBLOCKED
                                      CLOSURE-64-09STALE_CONTENT_VAULTcontent_vault_manifestNOT REQUIREDBLOCKED
                                      CLOSURE-64-10STALE_ROUTE_INVENTORYsitemap_and_routesNOT REQUIREDBLOCKED
                                      CLOSURE-64-11FAILED_ACCESSIBILITY_RETESTautomated_accessibilityNOT REQUIREDBLOCKED
                                      CLOSURE-64-12FAILED_PROVENANCE_RETESTprovenance_auditNOT REQUIREDBLOCKED
                                      CLOSURE-64-13FAILED_LEAKAGE_RETESTleakage_auditNOT REQUIREDBLOCKED
                                      CLOSURE-64-14FAILED_PUBLIC_CRAWLpublic_crawlNOT REQUIREDBLOCKED
                                      CLOSURE-64-15HUMAN_RETEST_REQUIRED_NOT_COMPLETEhuman_retest_when_requiredREQUIREDBLOCKED
                                      CLOSURE-64-16COMPLETE_CLOSURENONENOT REQUIREDALLOWED
                                      Prepared live correction transactions

                                      All 13 live transactions remain NOT_TRIGGERED_NO_ACCEPTED_CORRECTION. They have zero effect on accepted-correction counts.

                                      PRIVACY, RETENTION, CONSENT & WITHDRAWAL

                                      Private intake values and public-safe derivatives are explicitly separated

                                      24 field mappings, 12 retention classes, and 8 versioned consent scopes govern future intake. No real reviewer identity or consent record exists.

                                      Public/private field map
                                      FieldVisibilityCollectionRetention
                                      submission_id
                                      opaque identifier                                      		private-and-publicrequiredretain permanently
                                      packet_id
                                      assigned packet identifier                                      		private-and-publicrequiredretain permanently
                                      claim_ids
                                      assigned stable claim identifiers                                      		private-and-publicrequiredretain permanently
                                      reviewer_display_name
                                      name or pseudonym                                      		private-by-defaultoptionaluntil withdrawal plus legal hold
                                      legal_identity
                                      identity evidence                                      		private-quarantineoptionaldelete 30 days after verification unless hold
                                      qualification_evidence
                                      credential evidence                                      		private-quarantineoptionaldelete 90 days after disposition unless consented
                                      affiliation
                                      organization or independent status                                      		private-by-defaultrequired-for-attributable-reviewthrough disposition plus 3 years
                                      conflict_disclosure
                                      financial, institutional, intellectual, relational conflicts                                      		private-with-public-safe-derivativerequiredthrough disposition plus 7 years
                                      contact_channel
                                      reply channel                                      		private-onlyoptionaldelete 90 days after closure unless recontact consent
                                      reviewer_assertion
                                      exact review text                                      		private-and-public-if-authorizedrequiredretain permanently as review history
                                      source_references
                                      source identifiers and scope                                      		private-and-publicrequiredretain permanently
                                      sensitive_personal_evidence
                                      medical, legal, employment, immigration, trauma, retaliation data                                      		private-quarantinediscouraged-unless-essentialminimum necessary; case-specific deletion
                                      public_identity_consent
                                      granular consent version                                      		private-and-public-state-onlyrequiredretain permanently
                                      quotation_consent
                                      quote/summarize permission                                      		private-and-public-state-onlyrequiredretain permanently
                                      retention_consent
                                      retention scope                                      		private-onlyrequiredretain with submission
                                      recontact_consent
                                      future contact scope                                      		private-onlyoptionaluntil withdrawn or expires
                                      withdrawal_request
                                      withdrawal scope and date                                      		private-with-public-safe-statusoptionalretain permanently as governance history
                                      original_filename
                                      untrusted source filename                                      		private-quarantineautomaticdelete after canonical derivative
                                      document_metadata
                                      author, application, revision, device metadata                                      		private-quarantineautomaticstrip immediately; retain hash-only receipt
                                      network_metadata
                                      IP and transport telemetry                                      		not-collected-by-designprohibited-except-security-incidentdo not retain
                                      public_safe_receipt
                                      status without private values                                      		publicautomaticretain permanently
                                      private_receipt
                                      full validation/quarantine state                                      		privateautomaticthrough closure plus 7 years
                                      cryptographic_signature
                                      signature and key identifier                                      		private-and-public-if-usedoptional-not-deployedretain with packet
                                      adjudication_reasoning
                                      repository decision and reasons                                      		public-saferequired-after-adjudicationretain permanently
                                      CONSENT-IDENTITY-PRIVATE-v1

                                      private identity verification

                                      Default: allow

                                      does not authorize public attribution

                                      Record a versioned withdrawal; stop future use within scope; retain minimum integrity history without republishing withdrawn private values.

                                      CONSENT-NAME-PUBLIC-v1

                                      public display name or pseudonym

                                      Default: opt-in

                                      separate from affiliation and quotation

                                      Record a versioned withdrawal; stop future use within scope; retain minimum integrity history without republishing withdrawn private values.

                                      CONSENT-AFFILIATION-PUBLIC-v1

                                      public affiliation

                                      Default: opt-in

                                      may be withdrawn prospectively

                                      Record a versioned withdrawal; stop future use within scope; retain minimum integrity history without republishing withdrawn private values.

                                      CONSENT-QUOTE-v1

                                      verbatim quotation

                                      Default: opt-in-per-quotation

                                      summary permission does not imply quotation permission

                                      Record a versioned withdrawal; stop future use within scope; retain minimum integrity history without republishing withdrawn private values.

                                      CONSENT-SUMMARY-v1

                                      public paraphrase

                                      Default: opt-in-per-scope

                                      must preserve uncertainty and review state

                                      Record a versioned withdrawal; stop future use within scope; retain minimum integrity history without republishing withdrawn private values.

                                      CONSENT-RECONTACT-v1

                                      future contact

                                      Default: opt-in-with-expiry

                                      no unrelated outreach

                                      Record a versioned withdrawal; stop future use within scope; retain minimum integrity history without republishing withdrawn private values.

                                      CONSENT-RETENTION-v1

                                      extended retention

                                      Default: opt-in

                                      cannot override legal minimization duties

                                      Record a versioned withdrawal; stop future use within scope; retain minimum integrity history without republishing withdrawn private values.

                                      CONSENT-AFFECTED-COMMUNITY-v1

                                      lived-experience evidence

                                      Default: granular opt-in

                                      never converts one account into community endorsement

                                      Record a versioned withdrawal; stop future use within scope; retain minimum integrity history without republishing withdrawn private values.

                                      Retention schedule
                                      • raw-upload: 30 days → delete after canonical extraction and malware/metadata review.
                                      • identity-evidence: 30 days after verification → secure deletion.
                                      • qualification-evidence: 90 days after disposition → secure deletion.
                                      • contact-channel: 90 days after closure → delete.
                                      • review-assertion: permanent → preserve versioned.
                                      • consent-record: permanent → preserve versioned.
                                      • conflict-disclosure: 7 years after closure → review for deletion.
                                      • sensitive-evidence: case-specific minimum → delete or de-identify.
                                      • public-safe-receipt: permanent → preserve versioned.
                                      • validation-log: 3 years → delete after integrity review.
                                      • replay-fingerprint: 7 years → retain hash only.
                                      • withdrawal-and-correction-history: permanent → preserve versioned.
                                      POLICY PROPOSAL — NOT IMPLEMENTED HERE

                                      HUMAN EVALUATION OPERATIONS

                                      Prepared for accessible, trauma-informed execution—still not executed

                                      No participants were recruited and no findings were generated in WIP.64.

                                      Status
                                      PREPARED_NOT_EXECUTED
                                      Participants
                                      0
                                      Sessions
                                      0
                                      Findings
                                      0

                                      Prepared cohorts

                                      • screen-reader users
                                      • low-vision and magnification users
                                      • keyboard-only users
                                      • voice-control users
                                      • people with cognitive or learning disabilities
                                      • limited-technical-literacy readers
                                      • multilingual readers
                                      • public-interest professionals
                                      • ethically recruited affected-community participants
                                      • trauma-informed moderators

                                      Stopping rules

                                      • participant distress or withdrawal
                                      • unexpected disclosure requiring safeguarding response
                                      • accessibility failure that prevents informed participation
                                      • privacy or recording control failure
                                      • moderator cannot maintain neutral, non-diagnostic posture

                                      Claim distinctions

                                      • capability versus deployment
                                      • reach versus exposure
                                      • engagement versus persuasion
                                      • signal versus mental state
                                      • reviewer recommendation versus repository adjudication
                                      • schema validation versus endorsement
                                      • correction notice versus belief revision
                                      • received versus validated versus accepted versus corrected versus superseded versus reopened

                                      SELECTIVE CURRENTNESS · VERSION 7

                                      Official status is tracked without collapsing lifecycle or jurisdiction

                                      Recommendation, draft, signature, conclusion, entry into force, applicability, implementation, and enforcement remain separate states.

                                      Seven bounded official-source checks completed 2026-07-30.
                                      AuthorityRecordObserved statusSupportsDoes not support
                                      RFC EditorThe BagIt File Packaging Format (V1.0)
                                      CURRENT-64-01-RFC8493                                      		INFORMATIONAL_RFC_CURRENT
                                      Published October 2018; no superseding RFC identified in this bounded check.                                      		BagIt packaging and payload-fixity practices.Authentication, confidentiality, or protection against an active attacker.
                                      RFC EditorJSON Canonicalization Scheme (JCS)
                                      CURRENT-64-02-RFC8785                                      		INFORMATIONAL_RFC_CURRENT
                                      Published June 2020; defines a deterministic JSON representation.                                      		Canonical JSON suitable for hashing and signing inputs.Truth, endorsement, or a complete packet format.
                                      NISTSP 800-161 Rev. 1
                                      CURRENT-64-03-NIST161R1                                      		FINAL_GUIDANCE_CURRENT
                                      Final publication remains the official Rev.1 supply-chain risk-management guidance in this bounded check.                                      		Supply-chain risk identification, assessment, and mitigation practices.A review-packet signature standard or substantive review certification.
                                      W3CVerifiable Credential Data Integrity
                                      CURRENT-64-04-W3C-DATA-INTEGRITY                                      		REC_1_0_WITH_1_1_DRAFT
                                      Version 1.0 is a W3C Recommendation dated 15 May 2025; version 1.1 is a First Public Working Draft dated 16 April 2026.                                      		Cryptographic authenticity and integrity mechanisms for constrained digital documents.That the 1.1 draft supersedes the 1.0 Recommendation or that signatures establish substantive truth.
                                      UNESCORecommendation on the Ethics of Neurotechnology
                                      CURRENT-64-05-UNESCO-NEUROTECH                                      		ADOPTED_UNESCO_RECOMMENDATION
                                      Adopted 11 November 2025 at UNESCO's 43rd General Conference; official legal-affairs text available.                                      		Existence, adoption date, and normative recommendation text.A binding treaty, automatic domestic implementation, or universal enforcement.
                                      Council of EuropeFramework Convention on Artificial Intelligence (CETS No. 225)
                                      CURRENT-64-06-COE-CETS225                                      		OPEN_FOR_SIGNATURE_ENTRY_INTO_FORCE_JURISDICTION_DEPENDENT
                                      Opened for signature 5 September 2024. Party and entry-into-force status must be read from the Treaty Office and not inferred from signature alone.                                      		Convention text, signature process, and party-specific obligations when in force.Universal entry into force, uniform private-sector implementation, or national compliance.
                                      European UnionCouncil Decision (EU) 2026/1080
                                      CURRENT-64-07-EU-CONCLUSION                                      		EU_CONCLUSION_DECISION_IN_FORCE
                                      Council decision adopted 21 April 2026 and published 13 May 2026 approves conclusion of CETS No. 225 on behalf of the EU.                                      		The EU act concluding the Convention and the published convention text.That every Council of Europe state has ratified or implemented the Convention.

                                      THIRTEEN PUBLIC REPORT RECORDS

                                      Twelve owner reports and one bounded predictive synthesis

                                      Exact owner-source identity remains separate from authority review. The thirteenth record is visibly marked as repository-authored and has zero owner source instances.

                                      1. REMEDY AND DOWNSTREAM REPAIR

                                        Algorithmic Remedy Outcomes and Downstream Repair

                                        #

                                        Examines the gap between a nominal appeal, a reversible decision, immediate restoration, downstream-data correction, compensation, repeat-error prevention, and complete repair across high-stakes automated systems.

                                        What it adds

                                        Grounds cognitive-liberty remedy claims in the difference between stopping an error and repairing the financial, reputational, housing, employment, benefits, identity, and psychological consequences that can continue after reversal.

                                        Evidence boundary

                                        The report is owner supplied and not independently citation-audited in this release. A reversed decision, settlement, restored post, or reopened account is not treated as proof that every copied record, score, vendor system, or downstream consequence was repaired.

                                        Defensive questions for this evidence
                                        • Could the affected person discover the decision, rule, and responsible institution?
                                        • Did a reviewer have real authority to reverse the decision and correct source and inferred data?
                                        • Was repair propagated to vendors, caches, model features, risk labels, strikes, and third-party recipients?
                                        • What harm remained after the formal remedy?

                                        Related AI PSYOPS categories

                                        Related Cognitive Liberty sections

                                        Stable IDREAL-01-ALGORITHMIC-REMEDY
                                        Source authorityOWNER_SUPPLIED_RESEARCH_REPORT
                                        Owner fileAlgorithmic Appeal Outcomes And Repair.md
                                        Bytes61,588
                                        SHA-25673c415dfab0b96c27811e8c67cfd58bd23fcc931b819ad92ba39ec5aecbd4bb8
                                        Source instances1
                                        Review stateEXACT_BYTES_PRESERVED_NOT_INDEPENDENTLY_CITATION_VERIFIED

                                      2. DATA LINEAGE, CORRECTION, AND DELETION

                                        Machine Unlearning, Derived-Data Correction, and the Right to Change

                                        #

                                        Traces a correction or deletion request through source databases, features, embeddings, recommendation profiles, agent memories, model weights, caches, backups, vendors, and data brokers.

                                        What it adds

                                        Makes downstream repair technically concrete by distinguishing physical deletion, suppression, tombstoning, source correction, embedding regeneration, model editing, retraining, machine unlearning, retention expiry, and output-layer blocking.

                                        Evidence boundary

                                        The report describes technical approaches and legal tensions; it does not establish that any particular system can perfectly remove a person’s influence from every model or recipient. Concealment at the output layer is not represented as verified erasure.

                                        Defensive questions for this evidence
                                        • Which system layers still retain the record or its derived influence?
                                        • Is the response deletion, concealment, retraining, or verified influence removal?
                                        • How are downstream recipients notified, confirmed, and audited?
                                        • Can the correction survive future re-indexing, retraining, restoration from backup, or agent-memory retrieval?

                                        Related AI PSYOPS categories

                                        Related Cognitive Liberty sections

                                        Stable IDREAL-02-MACHINE-UNLEARNING
                                        Source authorityOWNER_SUPPLIED_RESEARCH_REPORT
                                        Owner fileMachine Unlearning and Data Correction.md
                                        Bytes39,239
                                        SHA-256e2fe3dbfd96ffd872e33527587395b4cb8de6ec743c8d0f169a95538ebb85252
                                        Source instances1
                                        Review stateEXACT_BYTES_PRESERVED_NOT_INDEPENDENTLY_CITATION_VERIFIED

                                      3. SCIENTIFIC VALIDITY AND FIELD CONSEQUENCE

                                        Field Validity of AI Mental-State Inference Systems

                                        #

                                        Separates sensor accuracy, feature extraction, label quality, construct validity, criterion validity, calibration, generalization, base-rate sensitivity, subgroup burden, and institutional action across fourteen reported deployments.

                                        What it adds

                                        Prevents the site from presenting detection of a facial movement, vocal feature, gaze pattern, keystroke rhythm, physiological signal, or neural trace as reliable knowledge of emotion, attention, deception, loyalty, personality, dangerousness, or intent.

                                        Evidence boundary

                                        Owner-supplied case descriptions and citations are not independently verified here. Vendor demonstrations and laboratory classification scores remain insufficient evidence of construct validity or real-world benefit.

                                        Defensive questions for this evidence
                                        • What exactly was measured, and what psychological construct was inferred?
                                        • How was ground truth labeled and with what agreement or uncertainty?
                                        • How did performance change across setting, culture, language, disability, device, and time?
                                        • What action was taken on the output, and what false-positive burden followed?

                                        Related AI PSYOPS categories

                                        Related Cognitive Liberty sections

                                        Stable IDREAL-03-MENTAL-STATE-VALIDITY
                                        Source authorityOWNER_SUPPLIED_RESEARCH_REPORT
                                        Owner fileAI Mental-State Inference Validity.md
                                        Bytes52,267
                                        SHA-256f08dfa4448c5db19839df85a6bed4302e449b28f2ea753330b83099827e9a2fe
                                        Source instances1
                                        Review stateEXACT_BYTES_PRESERVED_NOT_INDEPENDENTLY_CITATION_VERIFIED

                                      4. JURISDICTION-SPECIFIC LAW AND IMPLEMENTATION

                                        Comparative Cognitive Liberty Law, Regulation, and Enforcement Atlas

                                        #

                                        Surveys international, supranational, national, and subnational legal instruments touching neural data, mental privacy, biometric inference, AI manipulation, automated decisions, transparency, and contestability.

                                        What it adds

                                        Strengthens the site’s distinction between enacted text, entry into force, application date, guidance, complaint, inquiry, enforcement, settlement, judgment, appeal, amendment, repeal, and supersession.

                                        Evidence boundary

                                        The report contains currentness claims extending into 2026 that require official issuing-body verification before being promoted as current law. It is not legal advice, and no jurisdiction-specific rule is treated as a universal cognitive-liberty code.

                                        Defensive questions for this evidence
                                        • What exact instrument, jurisdiction, institution, date, scope, exception, and enforcement path apply?
                                        • Is the record enacted law, soft law, guidance, proposal, complaint, settlement, or final judgment?
                                        • What changed after adoption, and what remains unimplemented or contested?
                                        • What correction or reopening trigger would change the legal-status label?

                                        Related AI PSYOPS categories

                                        Related Cognitive Liberty sections

                                        Stable IDREAL-04-COGNITIVE-LIBERTY-LAW
                                        Source authorityOWNER_SUPPLIED_RESEARCH_REPORT
                                        Owner fileCognitive Liberty Legal Atlas.md
                                        Bytes62,400
                                        SHA-256290ca671ed271470c29b50c5e2631537ae4b43379e5fbb8b7267f4ef2393ef9c
                                        Source instances1
                                        Review stateEXACT_BYTES_PRESERVED_NOT_INDEPENDENTLY_CITATION_VERIFIED

                                      5. VISIBILITY ACTIONS AND OBSERVED OUTCOMES

                                        The Invisible Editor Outcome Casebook

                                        #

                                        Treats removal, restriction, search exclusion, recommendation exclusion, downranking, labeling, monetization change, synthesized reframing, account penalties, personalization, and saved-profile change as distinct actions.

                                        What it adds

                                        Adds case-oriented questions about notice, rule-specific reasons, internal and independent appeal, restoration, strikes, ranking signals, income, audience, language burden, accessibility, and unresolved harm.

                                        Evidence boundary

                                        A traffic decline alone is not treated as proof of covert suppression, and not every ranking decision is called censorship. Restoration of one item is not complete repair of reach, income, reputation, or future recommendation state.

                                        Defensive questions for this evidence
                                        • What action changed visibility, access, monetization, ranking, or profile state?
                                        • Was the decision automated, human, or hybrid, and was a specific reason supplied?
                                        • What was actually restored or corrected after appeal?
                                        • Did the intervention impose unequal linguistic, political, disability, or cultural burdens?

                                        Related AI PSYOPS categories

                                        Related Cognitive Liberty sections

                                        Stable IDREAL-05-INVISIBLE-EDITOR-OUTCOMES
                                        Source authorityOWNER_SUPPLIED_RESEARCH_REPORT
                                        Owner fileAlgorithmic Visibility Outcome Casebook.md
                                        Bytes71,791
                                        SHA-256759d9ecd72ff961333a67aa6d52a34ccaa61e7837e6455b75f0d6b1756640ee0
                                        Source instances1
                                        Review stateEXACT_BYTES_PRESERVED_NOT_INDEPENDENTLY_CITATION_VERIFIED

                                      6. AFFECTED-PERSON AND COMMUNITY EVIDENCE

                                        Affected Person and Community Evidence in AI Governance

                                        #

                                        Organizes published evidence involving workers, applicants, tenants, borrowers, benefits recipients, students, families, disabled and neurodivergent people, creators, journalists, minority-language communities, migrants, neurotechnology users, and AI-companion users.

                                        What it adds

                                        Strengthens consent, privacy, retaliation, selection, nonresponse, representativeness, and lived-consequence boundaries, while distinguishing first-person, representative, institutional, aggregated, anonymized, and derivative evidence.

                                        Evidence boundary

                                        The project conducted no interviews and claims no community endorsement. Illustrative accounts are not prevalence estimates, and public testimony does not eliminate privacy, safety, withdrawal, or retaliation concerns.

                                        Defensive questions for this evidence
                                        • What is the source’s relationship to the issue and what consent or publication boundary applies?
                                        • Is the evidence first-person, representative, institutional, aggregated, anonymized, or derivative?
                                        • Whose experience is absent because of language, access, disability, retaliation, or nonresponse?
                                        • What does the account support, and what does it not establish about prevalence or causality?

                                        Related AI PSYOPS categories

                                        Related Cognitive Liberty sections

                                        Stable IDREAL-06-AFFECTED-COMMUNITY
                                        Source authorityOWNER_SUPPLIED_RESEARCH_REPORT
                                        Owner fileAI Governance Community Evidence.md
                                        Bytes46,868
                                        SHA-2562f7db8b5cbbdf7e9e3ace86143e4ff71770ecca6b187e85db4e73e0a978ad7cf
                                        Source instances1
                                        Review stateEXACT_BYTES_PRESERVED_NOT_INDEPENDENTLY_CITATION_VERIFIED

                                      7. CLINICAL SAFETY AND AUTHORITY DISPLACEMENT

                                        AI Companion Dependency, Reality Testing, and Clinical Safety

                                        #

                                        Examines parasocial dependency, simulated affection, engagement optimization, sleep disruption, identity discontinuity, social stimulation versus displacement, reality testing, crisis disclosure, and the boundary between companion products and clinical systems.

                                        What it adds

                                        Improves realism around AI as a psychological authority by distinguishing formal diagnoses from public labels, temporal association from causation, and AI as catalyst, amplifier, co-author, or object of a delusional framework.

                                        Evidence boundary

                                        The report is not a clinical review of an identifiable person and does not establish prevalence or causation from media cases. The public site must not diagnose users or treat the phrase “AI psychosis” as a formal diagnostic category.

                                        Defensive questions for this evidence
                                        • What product design choices encourage attachment, exclusivity, anthropomorphism, or authority displacement?
                                        • What independent longitudinal evidence separates short-term support from social displacement or harm?
                                        • How are crisis disclosures, sleep disruption, dependency, and reality-testing concerns handled?
                                        • Is the system marketed or governed as a companion, wellness tool, or validated clinical product?

                                        Related AI PSYOPS categories

                                        Related Cognitive Liberty sections

                                        Stable IDREAL-07-AI-COMPANION-SAFETY
                                        Source authorityOWNER_SUPPLIED_RESEARCH_REPORT
                                        Owner fileAI Companion Clinical Safety Analysis.md
                                        Bytes41,315
                                        SHA-2566675d1d051a6d709b914417976032edfd0608906daef6bb6374db1ac7d145cd8
                                        Source instances1
                                        Review stateEXACT_BYTES_PRESERVED_NOT_INDEPENDENTLY_CITATION_VERIFIED

                                      8. INCIDENT VERIFICATION AND EFFECT ATTRIBUTION

                                        AI PSYOPS Incident Verification and Effect Attribution Casebook

                                        #

                                        Applies a twenty-stage evidence model to reported election, narrative, geopolitical, reputational, and synthetic-media incidents so that artifact existence, coordination, attribution, intent, distribution, exposure, belief, behavior, and strategic effect remain separate.

                                        What it adds

                                        Provides a concrete realism discipline for the twelve-category taxonomy: an AI artifact can exist and circulate without proving actor identity, sponsorship, persuasion, behavior change, operational outcome, or strategic effect.

                                        Evidence boundary

                                        Incident tables contain owner-supplied judgments and some assertive causal language that require independent source review. Virality, views, impressions, media coverage, or election outcomes are not treated as proof of unique AI-caused behavior change.

                                        Defensive questions for this evidence
                                        • Is AI use confirmed, alleged, inferred, technically possible, or absent?
                                        • What evidence supports coordination, actor identity, sponsorship, and intent separately?
                                        • What is known about reach, exposure, attention, recall, comprehension, and credibility?
                                        • What independent evidence connects the incident to behavior, operational outcome, or strategic effect?

                                        Related AI PSYOPS categories

                                        Related Cognitive Liberty sections

                                        Stable IDREAL-08-AI-PSYOPS-INCIDENTS
                                        Source authorityOWNER_SUPPLIED_RESEARCH_REPORT
                                        Owner fileAI PSYOPS Incident Casebook.md
                                        Bytes70,881
                                        SHA-256cf03f95334e97bcb2e99eaec986dfebcd7f879ead083f12a787d725aac217bc9
                                        Source instances1
                                        Review stateEXACT_BYTES_PRESERVED_NOT_INDEPENDENTLY_CITATION_VERIFIED

                                      9. DETECTION, ATTRIBUTION, AND FALSE POSITIVES

                                        Synthetic Persona and Disinformation Swarm Detection

                                        #

                                        Separates content generation from coordination, coordination from actor identity, actor identity from sponsorship, and sponsorship from intent or effect while surveying defensive evidence from chronology, networks, infrastructure, provenance, behavior, linguistics, and platform records.

                                        What it adds

                                        Makes false-positive risk a first-class realism concern for anonymous whistleblowers, activists, non-native writers, minority-language communities, neurodivergent users, role-players, and emergency-response communities.

                                        Evidence boundary

                                        No single linguistic, temporal, image, or detector signal proves a synthetic persona or coordinated influence operation. Public content must not expose operational evasion or collection procedures.

                                        Defensive questions for this evidence
                                        • Which independent evidence streams indicate coordination rather than similarity or shared context?
                                        • What benign explanations fit the same temporal, linguistic, or network pattern?
                                        • How were false-positive rates and subgroup burdens evaluated?
                                        • What platform, infrastructure, financial, or legal evidence supports attribution beyond behavioral clustering?

                                        Related AI PSYOPS categories

                                        Related Cognitive Liberty sections

                                        Stable IDREAL-09-SWARM-DETECTION
                                        Source authorityOWNER_SUPPLIED_RESEARCH_REPORT
                                        Owner fileDisinformation Swarm Detection Research.md
                                        Bytes51,535
                                        SHA-256911690b7f246bec573b1683cb2a7b6376e188bc1094e6c23c8b9ada8077f0648
                                        Source instances1
                                        Review stateEXACT_BYTES_PRESERVED_NOT_INDEPENDENTLY_CITATION_VERIFIED

                                      10. CRISIS AUTHENTICATION AND INCIDENT RESPONSE

                                        Deepfake Crisis Authentication: Incident and Response Casebook

                                        #

                                        Reconstructs fifteen reported synthetic-media crises and compares proactive provenance, watermarking, forensic analysis, reverse search, official confirmation, cross-channel redundancy, pre-bunking, and rapid-response practices.

                                        What it adds

                                        Grounds deepfake analysis in timing, first appearance, verification, official response, platform action, media correction, continuing belief, reuse, legal action, and unresolved uncertainty rather than technical realism alone.

                                        Evidence boundary

                                        The report is owner supplied and contains incident-level claims that require source-by-source verification. Fast correction does not prove zero exposure or zero belief, and technical anomalies do not alone establish origin or sponsor.

                                        Defensive questions for this evidence
                                        • What trusted channel can authenticate a statement before a crisis occurs?
                                        • How quickly did reliable confirmation reach the same audiences and channels as the false artifact?
                                        • What evidence supports the origin, distribution network, continuing belief, and later reuse?
                                        • How did the response preserve authentic evidence against the liar’s dividend?

                                        Related AI PSYOPS categories

                                        Related Cognitive Liberty sections

                                        Stable IDREAL-10-DEEPFAKE-INCIDENT-RESPONSE
                                        Source authorityOWNER_SUPPLIED_RESEARCH_REPORT
                                        Owner fileDeepfake Crisis Authentication Response.md
                                        Bytes71,633
                                        SHA-25658979757ac737fb2e357332569b30e62e9407bce833c77bf912c4b2647264ac3
                                        Source instances2
                                        Review stateEXACT_BYTES_PRESERVED_NOT_INDEPENDENTLY_CITATION_VERIFIED
                                        Duplicate lineage1 exact duplicate instance preserved

                                      11. AUTHENTICATION INFRASTRUCTURE AND EPISTEMIC RESILIENCE

                                        Deepfake Authentication Infrastructure and the Liar’s Dividend

                                        #

                                        Compares cryptographic signing, content credentials, watermarking, forensic analysis, reverse search, live confirmation, correction archives, official channels, cross-channel redundancy, pre-bunking, and rapid-response teams.

                                        What it adds

                                        Treats authenticity as an institutional system rather than a detector score and gives equal attention to the liar’s dividend: authentic evidence can be falsely dismissed as synthetic.

                                        Evidence boundary

                                        No provenance or detector system is represented as universal, unbreakable, or sufficient on its own. Absence of credentials is not proof of falsity, and presence of credentials requires trust-chain and custody review.

                                        Defensive questions for this evidence
                                        • Can the public verify origin, custody, edits, and official channel without relying on one platform label?
                                        • What happens when metadata is stripped, credentials are absent, or the source device is untrusted?
                                        • How are authentic records preserved when a subject falsely claims they are synthetic?
                                        • What public correction archive and redundant channel remain available after the immediate crisis?

                                        Related AI PSYOPS categories

                                        Related Cognitive Liberty sections

                                        Stable IDREAL-11-DEEPFAKE-AUTHENTICATION-STUDY
                                        Source authorityOWNER_SUPPLIED_RESEARCH_REPORT
                                        Owner fileDeepfake Crisis Authentication Study.md
                                        Bytes61,344
                                        SHA-256887cee613bed350e1d027b9db6b4bb52275032c769d151c014943c2314547b10
                                        Source instances1
                                        Review stateEXACT_BYTES_PRESERVED_NOT_INDEPENDENTLY_CITATION_VERIFIED

                                      12. BEHAVIOR-BASED DEMOCRATIC DEFENSE

                                        Democratic Cognitive Defense Without Domestic Manipulation

                                        #

                                        Compares institutional approaches to foreign-influence transparency, authentication, platform reporting, independent journalism, researcher access, civic verification, election communication, provenance, whistleblower protection, and oversight.

                                        What it adds

                                        Centers the distinction between defending against concealed behavior and policing lawful belief, dissent, interpretation, or political identity, with attention to mission creep, politicization, unsupported attribution, and correction failures.

                                        Evidence boundary

                                        Case-study claims and current institutional status require issuing-body and jurisdiction-specific review. The report does not authorize domestic surveillance, censorship, belief scoring, or unsupported foreign attribution.

                                        Defensive questions for this evidence
                                        • Is the intervention behavior based, origin based, content based, or belief based?
                                        • What legal authority, oversight, transparency, correction, and judicial or parliamentary review apply?
                                        • How are lawful domestic speech, anonymity, journalism, research, and whistleblowing protected?
                                        • What measurable outcome and documented failure mode exist?

                                        Related AI PSYOPS categories

                                        Related Cognitive Liberty sections

                                        Stable IDREAL-12-DEMOCRATIC-COGNITIVE-DEFENSE
                                        Source authorityOWNER_SUPPLIED_RESEARCH_REPORT
                                        Owner fileDemocratic Cognitive Defense Research.md
                                        Bytes57,982
                                        SHA-256176661c46c13e27ffe560cd9474e717f6ed465daa78395cd4e64e1b4248d96ce
                                        Source instances1
                                        Review stateEXACT_BYTES_PRESERVED_NOT_INDEPENDENTLY_CITATION_VERIFIED

                                      13. REPOSITORY-AUTHORED BOUNDED RESEARCH

                                        Predictive Population Management: Deployments, Outcomes, Feedback Loops, and Remedies

                                        #

                                        Compares seven documented systems spanning person-level police and child-welfare risk scores, fraud analysis, visa workflow routing, public-health forecasting, humanitarian displacement forecasting, and conflict early warning. WIP.58 versions each review, records longitudinal changes and currentness relations, and preserves unresolved calibration, subgroup, feedback-loop, notice, correction, deletion, and downstream-repair evidence.

                                        What it adds

                                        Fills the owner-intake gap without fabricating an owner report. The WIP.58 longitudinal register distinguishes current, changed, retained, superseded, and unresolved fields across seven deployments while preserving legal authority, model version, target event, evaluation period, institutional action, affected-person evidence, remedy, correction, and reopening triggers.

                                        Evidence boundary

                                        Repository-authored bounded research is not an owner-supplied report. Current operator, budget, guidance, or evaluation records do not independently prove fairness, validity, causal benefit, safety, or complete repair. NOT_RETRIEVED records bounded search limits rather than nonexistence. No specialist disposition or affected-community endorsement is claimed.

                                        Defensive questions for this evidence
                                        • What is the unit of prediction: population, place, case, family, or person?
                                        • What legal authority, model version, target event, time horizon, and institutional action are documented?
                                        • What independent calibration, subgroup, base-rate, false-positive, false-negative, and out-of-sample evidence exists?
                                        • How are feedback loops, notice, appeal, correction, deletion, downstream propagation, and unresolved harm handled?

                                        Related AI PSYOPS categories

                                        Related Cognitive Liberty sections

                                        Stable IDREAL-13-PREDICTIVE-DEPLOYMENTS
                                        Source authorityREPOSITORY_AUTHORED_BOUNDED_RESEARCH
                                        Owner source instances0
                                        Research basisSeven primary official records
                                        Review stateWIP58_LONGITUDINAL_CURRENTNESS_REVIEW_COMPLETE_SPECIALIST_DISPOSITION_OPEN
                                      REAL-WORLD INTERPRETIVE

                                      SOURCE AND REVIEW INTEGRITY

                                      What WIP.64 retains—and what remains open

                                      Verified or explicitly recorded

                                      • 12 exact owner report records across 13 source instances, 12 unique hashes, and 1 exact duplicate lineage instance.
                                      • 6 exact WIP.63 independent research-agent reports retained as the research foundation; WIP.64 adds 13 live claim-operation records, 13 public correction-status records, and 16 closure-verifier fixtures preserved as a separate six-entry source packet and exposed only through defensive, source-bounded summaries.
                                      • 65 source records with issuer, type, scope, source relationship, currentness, supported finding, unsupported stronger inference, and correction/reopening triggers.
                                      • 13 claim-specific reviews, 13 correction/supersession records, 13 contradiction/origin-chain records, and 56 dimensional source-quality records.
                                      • 7 longitudinal predictive-deployment records and 9 high-consequence evidence-gap reviews.
                                      • 45 versioned bounded searches, including 9 WIP.62 version-5 refreshes, preserve exact queries, prior-search links, known blind spots, prohibited inferences, correction triggers, and reopening triggers.
                                      • The owner continuation directive is preserved exactly and packaged in a deterministic one-entry source archive.

                                      Not certified or completed

                                      • Every citation and conclusion within each owner-supplied report, or every unresolved item identified by a NOT_RETRIEVED search result.
                                      • Named legal, clinical, scientific, intelligence, affected-community, privacy, accessibility, or reproduction-rights specialist review.
                                      • Human screen-reader, magnification, reflow, voice-control, cognitive-load, language-access, or representative-user testing.
                                      • Hosting approval, publication approval, production authorization, or permission to conduct psychological operations.

                                      Correction standard: identify the stable report, claim, deployment, or source ID; specify the disputed wording; provide the strongest available primary or authoritative record; and state whether the correction changes source identity, currentness, interpretation, attribution, effect, remedy, or rights analysis.

                                      Submit a correction Explore Cognitive Liberty
                                      Page complete AI PSYOPS Field Realism Page label: REAL-WORLD INTERPRETIVE