Claim, export conformance, and correction status

• Meta Oversight Board decision: Ayahuasca brew

  Source ID

  AUTH-03-META-OVERSIGHT-AYAHUASCA

  Class

  INDEPENDENT_PLATFORM_APPEAL_DECISION

  Relationship

  PRIMARY_APPELLATE_DECISION

  Independence

  PARTIAL_OR_UNRESOLVED

  Issuing body

  Meta Oversight Board

  Publication / access

  2021-01-28 / 2026-07-28

  Supports: Removal and appellate reversal · Notice and rule-clarity deficiencies · Restoration ordered after appeal

  Does not support: Restoration of prior reach or recommendation position · Correction of all account or ranking signals · Compensation or complete downstream repair

• Report of the Royal Commission into the Robodebt Scheme

  Source ID

  AUTH-04-ROBODEBT-ROYAL-COMMISSION

  Class

  OFFICIAL_INQUIRY_REPORT

  Relationship

  PRIMARY_OFFICIAL_INQUIRY

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  Australian Government — Royal Commission into the Robodebt Scheme

  Publication / access

  2023-07-07 / 2026-07-28

  Supports: Official inquiry findings and chronology · Institutional process failures · Published testimony and affected-person evidence within the inquiry record

  Does not support: That every individual harm or death had one exclusive cause · That all automated public-benefit systems share the same design · A universal remedy outcome

• New Robodebt class action settlement

  Source ID

  AUTH-35-ROBODEBT-2026-SETTLEMENT

  Class

  OFFICIAL_SETTLEMENT_IMPLEMENTATION_NOTICE

  Relationship

  OFFICIAL_NOTICE_OF_FEDERAL_COURT_APPROVED_SETTLEMENT

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  Services Australia / Federal Court-approved settlement information

  Publication / access

  2026-06-26 / 2026-07-28

  Supports: Settlement approval date · Additional compensation amount and relation to earlier refunds and debt zeroing · Appointment of a settlement administration process

  Does not support: Individual complete repair · Correction of every downstream record or inference · A merits judgment on every disputed issue

• Income Apportionment Resolution Scheme

  Source ID

  AUTH-36-ROBODEBT-INCOME-APPORTIONMENT-RESOLUTION

  Class

  OFFICIAL_REMEDY_PROGRAM_NOTICE

  Relationship

  PRIMARY_OPERATOR_REMEDY_PROGRAM_DESCRIPTION

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  Services Australia

  Publication / access

  2026-06-11 / 2026-07-28

  Supports: Existence, eligibility pathway, and dates of the resolution scheme · A separate remedy mechanism for eligible affected people

  Does not support: Take-up, accessibility, approval, or rejection rates · Complete compensation · Downstream data correction or repeat-error prevention

• Operation Myrtleford investigation report

  Source ID

  AUTH-37-ROBODEBT-NACC-MYRTLEFORD

  Class

  OFFICIAL_ANTI_CORRUPTION_INVESTIGATION_REPORT

  Relationship

  PRIMARY_INDEPENDENT_INVESTIGATION_REPORT

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  Australian National Anti-Corruption Commission

  Publication / access

  2026-03-11 / 2026-07-28

  Supports: The scope and findings of the NACC investigation · Specific accountability findings concerning events in 2015 and 2017

  Does not support: Criminal conviction · Complete institutional reform · Complete repair to affected people

• Parliamentary answers concerning SyRI data destruction and decision use

  Source ID

  AUTH-24-SYRI-DATA-DESTRUCTION

  Class

  OFFICIAL_PARLIAMENTARY_ANSWER

  Relationship

  PRIMARY_EXECUTIVE_ANSWER_TO_PARLIAMENT

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  Government of the Netherlands — House of Representatives records

  Publication / access

  2022-06-21 / 2026-07-28

  Supports: An official statement that SyRI-processed data had been destroyed by June 2022 · An official statement that responsible bodies did not make decisions based on SyRI-derived data · SyRI algorithms and criteria were not used in the named Detectie aan de Poort systems

  Does not support: Independent forensic verification of destruction · Deletion from every backup, derivative, vendor, or institutional system · Complete repair for communities subjected to SyRI projects

The first 104 entries preserve the exact WIP.70 baseline. The ninth entry per claim is a WIP.72 attribution-safe intake and source-quality event. No entry is an accepted correction or completed specialist disposition.

1. PRE-WIP58-RETAINED

   Effective date

   BEFORE_2026-07-28

   Change type

   RETAINED_HISTORICAL_WORDING

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   142588df2e74c96e99cec35f4dffd1c40e8e30b140164151346a932997ddfe0c

   Exact prior wording retained from the versioned correction ledger; this is not an accepted-correction event.

   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.

2. WIP58-CURRENTNESS-REFRESH

   Effective date

   2026-07-28

   Change type

   CURRENTNESS_REFRESH

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   74c79bb0c609b2371ebfef9d5af1ab873d982c07958db9eb503841b5d7f69a32

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

   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.

3. WIP65-CANONICAL

   Effective date

   2026-07-30

   Change type

   CANONICAL_OPERATION_NORMALIZATION_AND_CURRENTNESS_REFRESH

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   c053dbcc70fd5629d622f12034ce0b5080ec0f44153def96c4b647b1cd88a1bc

   WIP.65 canonicalized the live operation and public status; no accepted correction or real specialist review occurred.

   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.

4. WIP66-INTEROPERABILITY

   Effective date

   2026-07-31

   Change type

   CURRENTNESS_REFRESH_AND_MACHINE_EXPORT_INTEROPERABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   c053dbcc70fd5629d622f12034ce0b5080ec0f44153def96c4b647b1cd88a1bc

   WIP.66 adds deterministic machine export, public version history, queue and retest accountability, and currentness v9 without changing claim 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. Federal Court approval on 23 June 2026 of an additional Robodebt settlement documents compensation and remedy stages, not proof of complete downstream repair.

5. WIP67-EXPORT-V2-COMPATIBILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_V2_AND_EXTERNAL_ACKNOWLEDGMENT

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   c053dbcc70fd5629d622f12034ce0b5080ec0f44153def96c4b647b1cd88a1bc

   WIP.67 adds deterministic schema-v2 export, tested schema-v1 compatibility, immutable release-delta receipts, review-queue lifecycle fixtures, correction acknowledgment states, synthetic-only human-pilot dry runs, and currentness version 10 without changing this claim's 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. Federal Court approval on 23 June 2026 of an additional Robodebt settlement documents compensation and remedy stages, not proof of complete downstream repair.

6. WIP68-CONFORMANCE-AND-CORRECTION-OBSERVABILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_CONFORMANCE_AND_CORRECTION_OBSERVABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   c053dbcc70fd5629d622f12034ce0b5080ec0f44153def96c4b647b1cd88a1bc

   WIP.68 adds formal schema-v2/v1 conformance receipts, public correction observability, fail-closed queue-recovery fixtures, explicit external request and acknowledgment states, human execution-kit conformance, and currentness version 11 without changing this claim's 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. Federal Court approval on 23 June 2026 of an additional Robodebt settlement documents compensation and remedy stages, not proof of complete downstream repair.

7. WIP69-SCHEMA-EVOLUTION-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_SCHEMA_EVOLUTION_AND_EXTERNAL_DELIVERY_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   c053dbcc70fd5629d622f12034ce0b5080ec0f44153def96c4b647b1cd88a1bc

   WIP.69 adds additive schema migration/deprecation governance, prepared external delivery envelopes, queue SLA accountability, correction dependency graphs, and human-pilot authorization gates without changing claim 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. Federal Court approval on 23 June 2026 of an additional Robodebt settlement documents compensation and remedy stages, not proof of complete downstream repair.

8. WIP70-CONSUMER-MIGRATION-AND-REHEARSAL-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_CONSUMER_MIGRATION_REHEARSAL_AND_CORRECTION_EXECUTION_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   c053dbcc70fd5629d622f12034ce0b5080ec0f44153def96c4b647b1cd88a1bc

   WIP.70 adds consumer-migration evidence classification, synthetic external-dispatch rehearsal, correction-graph execution simulation, queue-operator accountability, and human-pilot preauthorization without changing claim 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. Federal Court approval on 23 June 2026 of an additional Robodebt settlement documents compensation and remedy stages, not proof of complete downstream repair.

9. WIP.71

   Effective date

   2026-08-01

   Change type

   ACCOUNTABILITY_AND_SOURCE_QUALITY_APPEND_ONLY_NO_WORDING_CHANGE

   Accepted correction

   NO

   Repository adjudication

   NONE

   Wording SHA-256

   c053dbcc70fd5629d622f12034ce0b5080ec0f44153def96c4b647b1cd88a1bc

   WIP.71 adds attribution-safe intake, external-receipt verifier parity, target-level correction provenance, queue audit-chain controls, human-pilot governance rehearsals, source-quality panels, and currentness v14 without changing claim wording or real authority counts.

   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.

10. WIP72-EVIDENCE-QUARANTINE-AND-DECISION-CONTROL-TRACEABILITY

    Effective date

    2026-08-01

    Change type

    EVIDENCE_GOVERNANCE_AND_SOURCE_TRACEABILITY_NO_WORDING_CHANGE

    Accepted correction

    NO

    Repository adjudication

    NOT_COMPLETED

    Wording SHA-256

    c053dbcc70fd5629d622f12034ce0b5080ec0f44153def96c4b647b1cd88a1bc

    WIP.72 adds fail-closed evidence quarantine, receipt reconciliation, correction conflict handling, human-pilot preparation, and AI kill-chain citation traceability without changing this claim 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. Federal Court approval on 23 June 2026 of an additional Robodebt settlement documents compensation and remedy stages, not proof of complete downstream repair.

• NIST glossary entry: Machine Unlearning

  Source ID

  AUTH-08-NIST-MACHINE-UNLEARNING

  Class

  OFFICIAL_TECHNICAL_DEFINITION

  Relationship

  PRIMARY_OFFICIAL_DEFINITION

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  U.S. National Institute of Standards and Technology

  Publication / access

  2025-01-01 / 2026-07-28

  Supports: A bounded authoritative definition of machine unlearning · The distinction between training-data influence removal and simple record deletion

  Does not support: That every unlearning method is exact · That output blocking removes model influence · Legal compliance for any specific implementation

• Machine Unlearning (SISA training architecture)

  Source ID

  AUTH-09-SISA-UNLEARNING

  Class

  PRIMARY_RESEARCH_PREPRINT

  Relationship

  PRIMARY_RESEARCH_ARTICLE

  Independence

  PARTIAL_OR_UNRESOLVED

  Issuing body

  Peer-reviewed research authors / arXiv record

  Publication / access

  2019-12-09 / 2026-07-28

  Supports: A concrete architecture for efficient data-removal workflows · Technical distinction between full retraining and bounded retraining

  Does not support: Perfect unlearning for every model class · Legal sufficiency · Removal from downstream vendors, logs, or backups

• New framework for auditing machine unlearning

  Source ID

  AUTH-40-GOOGLE-UNLEARNING-AUDIT-2026

  Class

  PRIMARY_RESEARCH_ANNOUNCEMENT

  Relationship

  PRIMARY_RESEARCH_SUMMARY

  Independence

  PARTIAL_OR_UNRESOLVED

  Issuing body

  Google Research

  Publication / access

  2026-06-10 / 2026-07-28

  Supports: Existence and described scope of a new auditing framework · The distinction between performing and auditing unlearning · The need for statistical evidence rather than UI-level deletion claims

  Does not support: Exact deletion in deployed foundation models · Production deployment across vendors · Elimination of every memorized or inferential effect

• Unlearning or Obfuscating? Jogging the Memory of Unlearned LLMs via Benign Relearning

  Source ID

  AUTH-41-ICLR-BENIGN-RELEARNING

  Class

  PEER_REVIEWED_MACHINE_UNLEARNING_EVALUATION

  Relationship

  PEER_REVIEWED_EMPIRICAL_SECURITY_EVALUATION

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  ICLR 2025 peer-reviewed research

  Publication / access

  2025-04-01 / 2026-07-28

  Supports: Reversibility risk for evaluated methods and datasets · A distinction between obfuscation and robust forgetting · Need for adversarial evaluation of unlearning claims

  Does not support: Failure of every unlearning method · Production-vendor deletion outcomes · Exact recovery of every target item

• Auditing Machine Unlearning: A Systematic Research on Whether Existing Algorithms Truly Forget

  Source ID

  AUTH-52-UNLEARNING-SYSTEMATIC-AUDIT-2026

  Class

  ACADEMIC_PREPRINT_SYSTEMATIC_AUDIT

  Relationship

  INDEPENDENT_METHODS_AND_LIMITATIONS_STUDY

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  Independent academic authors

  Publication / access

  2026-06-15 / 2026-07-28

  Supports: Unlearning auditability remains an open technical problem · Failed extraction or membership tests do not automatically prove exact forgetting · Residual information leakage must be evaluated separately from ordinary output suppression

  Does not support: A universal failure of every unlearning method · Exact behavior of any named production foundation model · Legal compliance in a specific jurisdiction

• Adversarial Machine Learning: A Taxonomy and Terminology of Attacks and Mitigations

  Source ID

  AUTH-39-NIST-AML-UNLEARNING

  Class

  PRIMARY_GOVERNMENT_TECHNICAL_GUIDANCE

  Relationship

  PRIMARY_GOVERNMENT_GUIDANCE

  Independence

  NOT_RECORDED

  Issuing body

  U.S. National Institute of Standards and Technology

  Publication / access

  NOT_RECORDED / NOT_RECORDED

  Supports: Recognizes machine unlearning and distinguishes exact and approximate approaches.

  Does not support: That any named production model has achieved exact audited unlearning or legal compliance.

• Auditing Machine Unlearning: A Systematic Research on Whether Models Truly Forget

  Source ID

  AUTH-40-UNLEARNING-AUDIT-2026

  Class

  PRIMARY_RESEARCH_PREPRINT

  Relationship

  INDEPENDENT_RESEARCH_PREPRINT

  Independence

  NOT_RECORDED

  Issuing body

  Ye et al.

  Publication / access

  NOT_RECORDED / NOT_RECORDED

  Supports: Documents that auditing whether unlearning truly erases influence remains an open technical challenge and evaluates multiple methods.

  Does not support: Peer-reviewed consensus, universal failure, or production deployment validity.

The first 104 entries preserve the exact WIP.70 baseline. The ninth entry per claim is a WIP.72 attribution-safe intake and source-quality event. No entry is an accepted correction or completed specialist disposition.

1. PRE-WIP58-RETAINED

   Effective date

   BEFORE_2026-07-28

   Change type

   RETAINED_HISTORICAL_WORDING

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   f4991458d196251f96a9c52d16dca76418f314e893860424910a8d75d683be49

   Exact prior wording retained from the versioned correction ledger; this is not an accepted-correction event.

   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.

2. WIP58-CURRENTNESS-REFRESH

   Effective date

   2026-07-28

   Change type

   NARROWING

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   bc3abebf2c96c63a295b6020879d30f01729c221315aa58465792b0f7d986fb9

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

   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.

3. WIP65-CANONICAL

   Effective date

   2026-07-30

   Change type

   CANONICAL_OPERATION_NORMALIZATION_AND_CURRENTNESS_REFRESH

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   68276f0748259f644274db399fd2f54e3fdea5b981e7c18081c971807332e9b0

   WIP.65 canonicalized the live operation and public status; no accepted correction or real specialist review occurred.

   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.

4. WIP66-INTEROPERABILITY

   Effective date

   2026-07-31

   Change type

   CURRENTNESS_REFRESH_AND_MACHINE_EXPORT_INTEROPERABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   68276f0748259f644274db399fd2f54e3fdea5b981e7c18081c971807332e9b0

   WIP.66 adds deterministic machine export, public version history, queue and retest accountability, and currentness v9 without changing claim wording.

   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.

5. WIP67-EXPORT-V2-COMPATIBILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_V2_AND_EXTERNAL_ACKNOWLEDGMENT

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   68276f0748259f644274db399fd2f54e3fdea5b981e7c18081c971807332e9b0

   WIP.67 adds deterministic schema-v2 export, tested schema-v1 compatibility, immutable release-delta receipts, review-queue lifecycle fixtures, correction acknowledgment states, synthetic-only human-pilot dry runs, and currentness version 10 without changing this claim's wording.

   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.

6. WIP68-CONFORMANCE-AND-CORRECTION-OBSERVABILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_CONFORMANCE_AND_CORRECTION_OBSERVABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   68276f0748259f644274db399fd2f54e3fdea5b981e7c18081c971807332e9b0

   WIP.68 adds formal schema-v2/v1 conformance receipts, public correction observability, fail-closed queue-recovery fixtures, explicit external request and acknowledgment states, human execution-kit conformance, and currentness version 11 without changing this claim's wording.

   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.

7. WIP69-SCHEMA-EVOLUTION-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_SCHEMA_EVOLUTION_AND_EXTERNAL_DELIVERY_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   68276f0748259f644274db399fd2f54e3fdea5b981e7c18081c971807332e9b0

   WIP.69 adds additive schema migration/deprecation governance, prepared external delivery envelopes, queue SLA accountability, correction dependency graphs, and human-pilot authorization gates without changing claim wording.

   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.

8. WIP70-CONSUMER-MIGRATION-AND-REHEARSAL-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_CONSUMER_MIGRATION_REHEARSAL_AND_CORRECTION_EXECUTION_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   68276f0748259f644274db399fd2f54e3fdea5b981e7c18081c971807332e9b0

   WIP.70 adds consumer-migration evidence classification, synthetic external-dispatch rehearsal, correction-graph execution simulation, queue-operator accountability, and human-pilot preauthorization without changing claim wording.

   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.

9. WIP.71

   Effective date

   2026-08-01

   Change type

   ACCOUNTABILITY_AND_SOURCE_QUALITY_APPEND_ONLY_NO_WORDING_CHANGE

   Accepted correction

   NO

   Repository adjudication

   NONE

   Wording SHA-256

   68276f0748259f644274db399fd2f54e3fdea5b981e7c18081c971807332e9b0

   WIP.71 adds attribution-safe intake, external-receipt verifier parity, target-level correction provenance, queue audit-chain controls, human-pilot governance rehearsals, source-quality panels, and currentness v14 without changing claim wording or real authority counts.

   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.

10. WIP72-EVIDENCE-QUARANTINE-AND-DECISION-CONTROL-TRACEABILITY

    Effective date

    2026-08-01

    Change type

    EVIDENCE_GOVERNANCE_AND_SOURCE_TRACEABILITY_NO_WORDING_CHANGE

    Accepted correction

    NO

    Repository adjudication

    NOT_COMPLETED

    Wording SHA-256

    68276f0748259f644274db399fd2f54e3fdea5b981e7c18081c971807332e9b0

    WIP.72 adds fail-closed evidence quarantine, receipt reconciliation, correction conflict handling, human-pilot preparation, and AI kill-chain citation traceability without changing this claim wording.

    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.

• Emotional Expressions Reconsidered: Challenges to Inferring Emotion From Human Facial Movements

  Source ID

  AUTH-10-BARRETT-FACIAL-MOVEMENTS

  Class

  PEER_REVIEWED_SCIENTIFIC_REVIEW

  Relationship

  PRIMARY_PEER_REVIEWED_REVIEW

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  Association for Psychological Science authors / Psychological Science in the Public Interest

  Publication / access

  2019-07-17 / 2026-07-28

  Supports: Construct-validity caution for facial emotion inference · Context, culture, and individual variation

  Does not support: That facial movement detection itself is impossible · That every affective-computing application has zero utility · Validity of other modalities without separate evidence

• Regulation (EU) 2024/1689 — Artificial Intelligence Act

  Source ID

  AUTH-05-EU-AI-ACT

  Class

  ENACTED_LEGISLATION

  Relationship

  PRIMARY_LEGAL_TEXT

  Independence

  PARTIAL_OR_UNRESOLVED

  Issuing body

  European Union

  Publication / access

  2024-07-12 / 2026-07-28

  Supports: Exact enacted text and jurisdictional scope · Article 5 prohibition categories and exceptions · Application schedule stated in the Regulation

  Does not support: A universal cognitive-liberty code · Uniform implementation outside the EU · A scientific finding that all emotion inference is invalid

• Automated Emotion Recognition in the Workplace

  Source ID

  AUTH-42-ACM-WORKPLACE-EMOTION-RECOGNITION

  Class

  PEER_REVIEWED_SYSTEMS_AND_WORKPLACE_ANALYSIS

  Relationship

  PEER_REVIEWED_ANALYSIS_OF_PROPOSED_SYSTEMS

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  ACM / peer-reviewed HCI research

  Publication / access

  2023-01-01 / 2026-07-28

  Supports: Documented design proposals and claimed workplace uses · The gap between sensed proxies, inferred labels, and institutional action · Privacy, labor, and validity concerns

  Does not support: Independent construct validity for a specified deployed system · Clinical knowledge of employee emotion · Population-wide performance

• Emotion AI in Job Interviews: Injustice, Emotional Labor, and Design Implications

  Source ID

  AUTH-43-ACM-JOB-INTERVIEW-EMOTION-AI

  Class

  PEER_REVIEWED_USER_STUDY

  Relationship

  PEER_REVIEWED_AFFECTED_USER_STUDY

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  ACM / peer-reviewed HCI research

  Publication / access

  2025-06-23 / 2026-07-28

  Supports: Affected-user perceptions of procedural justice in the studied interview context · Potential emotional labor and fairness burdens

  Does not support: Construct validity of the emotion score · Hiring outcome benefit or predictive validity · Generalization to every hiring system

The first 104 entries preserve the exact WIP.70 baseline. The ninth entry per claim is a WIP.72 attribution-safe intake and source-quality event. No entry is an accepted correction or completed specialist disposition.

1. PRE-WIP58-RETAINED

   Effective date

   BEFORE_2026-07-28

   Change type

   RETAINED_HISTORICAL_WORDING

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   a0eb60715b9f97c8a4f55bab5268536208a4def639ceb4e71e8b0e0d94948c11

   Exact prior wording retained from the versioned correction ledger; this is not an accepted-correction event.

   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.

2. WIP58-CURRENTNESS-REFRESH

   Effective date

   2026-07-28

   Change type

   CLARIFICATION

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   8ab000762a883875380f4d8d7d02dd9693161799449b51a5be32c3724613d4c5

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

   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.

3. WIP65-CANONICAL

   Effective date

   2026-07-30

   Change type

   CANONICAL_OPERATION_NORMALIZATION_AND_CURRENTNESS_REFRESH

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   8ab000762a883875380f4d8d7d02dd9693161799449b51a5be32c3724613d4c5

   WIP.65 canonicalized the live operation and public status; no accepted correction or real specialist review occurred.

   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.

4. WIP66-INTEROPERABILITY

   Effective date

   2026-07-31

   Change type

   CURRENTNESS_REFRESH_AND_MACHINE_EXPORT_INTEROPERABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   8ab000762a883875380f4d8d7d02dd9693161799449b51a5be32c3724613d4c5

   WIP.66 adds deterministic machine export, public version history, queue and retest accountability, and currentness v9 without changing claim 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.

5. WIP67-EXPORT-V2-COMPATIBILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_V2_AND_EXTERNAL_ACKNOWLEDGMENT

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   8ab000762a883875380f4d8d7d02dd9693161799449b51a5be32c3724613d4c5

   WIP.67 adds deterministic schema-v2 export, tested schema-v1 compatibility, immutable release-delta receipts, review-queue lifecycle fixtures, correction acknowledgment states, synthetic-only human-pilot dry runs, and currentness version 10 without changing this claim's 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.

6. WIP68-CONFORMANCE-AND-CORRECTION-OBSERVABILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_CONFORMANCE_AND_CORRECTION_OBSERVABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   8ab000762a883875380f4d8d7d02dd9693161799449b51a5be32c3724613d4c5

   WIP.68 adds formal schema-v2/v1 conformance receipts, public correction observability, fail-closed queue-recovery fixtures, explicit external request and acknowledgment states, human execution-kit conformance, and currentness version 11 without changing this claim's 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.

7. WIP69-SCHEMA-EVOLUTION-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_SCHEMA_EVOLUTION_AND_EXTERNAL_DELIVERY_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   8ab000762a883875380f4d8d7d02dd9693161799449b51a5be32c3724613d4c5

   WIP.69 adds additive schema migration/deprecation governance, prepared external delivery envelopes, queue SLA accountability, correction dependency graphs, and human-pilot authorization gates without changing claim 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.

8. WIP70-CONSUMER-MIGRATION-AND-REHEARSAL-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_CONSUMER_MIGRATION_REHEARSAL_AND_CORRECTION_EXECUTION_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   8ab000762a883875380f4d8d7d02dd9693161799449b51a5be32c3724613d4c5

   WIP.70 adds consumer-migration evidence classification, synthetic external-dispatch rehearsal, correction-graph execution simulation, queue-operator accountability, and human-pilot preauthorization without changing claim 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.

9. WIP.71

   Effective date

   2026-08-01

   Change type

   ACCOUNTABILITY_AND_SOURCE_QUALITY_APPEND_ONLY_NO_WORDING_CHANGE

   Accepted correction

   NO

   Repository adjudication

   NONE

   Wording SHA-256

   8ab000762a883875380f4d8d7d02dd9693161799449b51a5be32c3724613d4c5

   WIP.71 adds attribution-safe intake, external-receipt verifier parity, target-level correction provenance, queue audit-chain controls, human-pilot governance rehearsals, source-quality panels, and currentness v14 without changing claim wording or real authority counts.

   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.

10. WIP72-EVIDENCE-QUARANTINE-AND-DECISION-CONTROL-TRACEABILITY

    Effective date

    2026-08-01

    Change type

    EVIDENCE_GOVERNANCE_AND_SOURCE_TRACEABILITY_NO_WORDING_CHANGE

    Accepted correction

    NO

    Repository adjudication

    NOT_COMPLETED

    Wording SHA-256

    8ab000762a883875380f4d8d7d02dd9693161799449b51a5be32c3724613d4c5

    WIP.72 adds fail-closed evidence quarantine, receipt reconciliation, correction conflict handling, human-pilot preparation, and AI kill-chain citation traceability without changing this claim 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.

• Regulation (EU) 2024/1689 — Artificial Intelligence Act

  Source ID

  AUTH-05-EU-AI-ACT

  Class

  ENACTED_LEGISLATION

  Relationship

  PRIMARY_LEGAL_TEXT

  Independence

  PARTIAL_OR_UNRESOLVED

  Issuing body

  European Union

  Publication / access

  2024-07-12 / 2026-07-28

  Supports: Exact enacted text and jurisdictional scope · Article 5 prohibition categories and exceptions · Application schedule stated in the Regulation

  Does not support: A universal cognitive-liberty code · Uniform implementation outside the EU · A scientific finding that all emotion inference is invalid

• Colorado HB24-1058 — Protect Privacy of Biological Data

  Source ID

  AUTH-06-COLORADO-NEURAL-DATA

  Class

  ENACTED_STATE_LEGISLATION

  Relationship

  PRIMARY_LEGISLATIVE_RECORD

  Independence

  PARTIAL_OR_UNRESOLVED

  Issuing body

  Colorado General Assembly

  Publication / access

  2024-04-17 / 2026-07-28

  Supports: Existence and official status of the Colorado measure · Jurisdiction-bounded neural-data protection

  Does not support: A federal U.S. neurorights regime · Universal mental-privacy protection · How every neural-data product will be enforced

The first 104 entries preserve the exact WIP.70 baseline. The ninth entry per claim is a WIP.72 attribution-safe intake and source-quality event. No entry is an accepted correction or completed specialist disposition.

1. PRE-WIP58-RETAINED

   Effective date

   BEFORE_2026-07-28

   Change type

   RETAINED_HISTORICAL_WORDING

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   c7dec817644a7ab25ca69d5dc0723ac789c951391767623157dc97f03cb80f56

   Exact prior wording retained from the versioned correction ledger; this is not an accepted-correction event.

   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.

2. WIP58-CURRENTNESS-REFRESH

   Effective date

   2026-07-28

   Change type

   NO_SUBSTANTIVE_CHANGE

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   5dd0fc5f0305048f8f836ef0a56162a6e0831bf4ebbef1065e429c55166cc3ff

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

   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.

3. WIP65-CANONICAL

   Effective date

   2026-07-30

   Change type

   CANONICAL_OPERATION_NORMALIZATION_AND_CURRENTNESS_REFRESH

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   5dd0fc5f0305048f8f836ef0a56162a6e0831bf4ebbef1065e429c55166cc3ff

   WIP.65 canonicalized the live operation and public status; no accepted correction or real specialist review occurred.

   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.

4. WIP66-INTEROPERABILITY

   Effective date

   2026-07-31

   Change type

   CURRENTNESS_REFRESH_AND_MACHINE_EXPORT_INTEROPERABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   5dd0fc5f0305048f8f836ef0a56162a6e0831bf4ebbef1065e429c55166cc3ff

   WIP.66 adds deterministic machine export, public version history, queue and retest accountability, and currentness v9 without changing claim 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.

5. WIP67-EXPORT-V2-COMPATIBILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_V2_AND_EXTERNAL_ACKNOWLEDGMENT

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   5dd0fc5f0305048f8f836ef0a56162a6e0831bf4ebbef1065e429c55166cc3ff

   WIP.67 adds deterministic schema-v2 export, tested schema-v1 compatibility, immutable release-delta receipts, review-queue lifecycle fixtures, correction acknowledgment states, synthetic-only human-pilot dry runs, and currentness version 10 without changing this claim's 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.

6. WIP68-CONFORMANCE-AND-CORRECTION-OBSERVABILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_CONFORMANCE_AND_CORRECTION_OBSERVABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   5dd0fc5f0305048f8f836ef0a56162a6e0831bf4ebbef1065e429c55166cc3ff

   WIP.68 adds formal schema-v2/v1 conformance receipts, public correction observability, fail-closed queue-recovery fixtures, explicit external request and acknowledgment states, human execution-kit conformance, and currentness version 11 without changing this claim's 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.

7. WIP69-SCHEMA-EVOLUTION-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_SCHEMA_EVOLUTION_AND_EXTERNAL_DELIVERY_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   5dd0fc5f0305048f8f836ef0a56162a6e0831bf4ebbef1065e429c55166cc3ff

   WIP.69 adds additive schema migration/deprecation governance, prepared external delivery envelopes, queue SLA accountability, correction dependency graphs, and human-pilot authorization gates without changing claim 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.

8. WIP70-CONSUMER-MIGRATION-AND-REHEARSAL-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_CONSUMER_MIGRATION_REHEARSAL_AND_CORRECTION_EXECUTION_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   5dd0fc5f0305048f8f836ef0a56162a6e0831bf4ebbef1065e429c55166cc3ff

   WIP.70 adds consumer-migration evidence classification, synthetic external-dispatch rehearsal, correction-graph execution simulation, queue-operator accountability, and human-pilot preauthorization without changing claim 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.

9. WIP.71

   Effective date

   2026-08-01

   Change type

   ACCOUNTABILITY_AND_SOURCE_QUALITY_APPEND_ONLY_NO_WORDING_CHANGE

   Accepted correction

   NO

   Repository adjudication

   NONE

   Wording SHA-256

   5dd0fc5f0305048f8f836ef0a56162a6e0831bf4ebbef1065e429c55166cc3ff

   WIP.71 adds attribution-safe intake, external-receipt verifier parity, target-level correction provenance, queue audit-chain controls, human-pilot governance rehearsals, source-quality panels, and currentness v14 without changing claim wording or real authority counts.

   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.

10. WIP72-EVIDENCE-QUARANTINE-AND-DECISION-CONTROL-TRACEABILITY

    Effective date

    2026-08-01

    Change type

    EVIDENCE_GOVERNANCE_AND_SOURCE_TRACEABILITY_NO_WORDING_CHANGE

    Accepted correction

    NO

    Repository adjudication

    NOT_COMPLETED

    Wording SHA-256

    5dd0fc5f0305048f8f836ef0a56162a6e0831bf4ebbef1065e429c55166cc3ff

    WIP.72 adds fail-closed evidence quarantine, receipt reconciliation, correction conflict handling, human-pilot preparation, and AI kill-chain citation traceability without changing this claim 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.

• Meta Oversight Board decision: Ayahuasca brew

  Source ID

  AUTH-03-META-OVERSIGHT-AYAHUASCA

  Class

  INDEPENDENT_PLATFORM_APPEAL_DECISION

  Relationship

  PRIMARY_APPELLATE_DECISION

  Independence

  PARTIAL_OR_UNRESOLVED

  Issuing body

  Meta Oversight Board

  Publication / access

  2021-01-28 / 2026-07-28

  Supports: Removal and appellate reversal · Notice and rule-clarity deficiencies · Restoration ordered after appeal

  Does not support: Restoration of prior reach or recommendation position · Correction of all account or ranking signals · Compensation or complete downstream repair

The first 104 entries preserve the exact WIP.70 baseline. The ninth entry per claim is a WIP.72 attribution-safe intake and source-quality event. No entry is an accepted correction or completed specialist disposition.

1. PRE-WIP58-RETAINED

   Effective date

   BEFORE_2026-07-28

   Change type

   RETAINED_HISTORICAL_WORDING

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   df612ae922150518bc448a957d871e697a927e5becccee7c5e92455052bd5265

   Exact prior wording retained from the versioned correction ledger; this is not an accepted-correction event.

   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.

2. WIP58-CURRENTNESS-REFRESH

   Effective date

   2026-07-28

   Change type

   CLARIFICATION

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cb2fa8aaa0207a90deffb4195a0d359429f28c4c4ab1e2d263c987dd7f4d1a62

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

   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.

3. WIP65-CANONICAL

   Effective date

   2026-07-30

   Change type

   CANONICAL_OPERATION_NORMALIZATION_AND_CURRENTNESS_REFRESH

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cb2fa8aaa0207a90deffb4195a0d359429f28c4c4ab1e2d263c987dd7f4d1a62

   WIP.65 canonicalized the live operation and public status; no accepted correction or real specialist review occurred.

   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.

4. WIP66-INTEROPERABILITY

   Effective date

   2026-07-31

   Change type

   CURRENTNESS_REFRESH_AND_MACHINE_EXPORT_INTEROPERABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cb2fa8aaa0207a90deffb4195a0d359429f28c4c4ab1e2d263c987dd7f4d1a62

   WIP.66 adds deterministic machine export, public version history, queue and retest accountability, and currentness v9 without changing claim 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.

5. WIP67-EXPORT-V2-COMPATIBILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_V2_AND_EXTERNAL_ACKNOWLEDGMENT

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cb2fa8aaa0207a90deffb4195a0d359429f28c4c4ab1e2d263c987dd7f4d1a62

   WIP.67 adds deterministic schema-v2 export, tested schema-v1 compatibility, immutable release-delta receipts, review-queue lifecycle fixtures, correction acknowledgment states, synthetic-only human-pilot dry runs, and currentness version 10 without changing this claim's 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.

6. WIP68-CONFORMANCE-AND-CORRECTION-OBSERVABILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_CONFORMANCE_AND_CORRECTION_OBSERVABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cb2fa8aaa0207a90deffb4195a0d359429f28c4c4ab1e2d263c987dd7f4d1a62

   WIP.68 adds formal schema-v2/v1 conformance receipts, public correction observability, fail-closed queue-recovery fixtures, explicit external request and acknowledgment states, human execution-kit conformance, and currentness version 11 without changing this claim's 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.

7. WIP69-SCHEMA-EVOLUTION-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_SCHEMA_EVOLUTION_AND_EXTERNAL_DELIVERY_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cb2fa8aaa0207a90deffb4195a0d359429f28c4c4ab1e2d263c987dd7f4d1a62

   WIP.69 adds additive schema migration/deprecation governance, prepared external delivery envelopes, queue SLA accountability, correction dependency graphs, and human-pilot authorization gates without changing claim 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.

8. WIP70-CONSUMER-MIGRATION-AND-REHEARSAL-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_CONSUMER_MIGRATION_REHEARSAL_AND_CORRECTION_EXECUTION_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cb2fa8aaa0207a90deffb4195a0d359429f28c4c4ab1e2d263c987dd7f4d1a62

   WIP.70 adds consumer-migration evidence classification, synthetic external-dispatch rehearsal, correction-graph execution simulation, queue-operator accountability, and human-pilot preauthorization without changing claim 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.

9. WIP.71

   Effective date

   2026-08-01

   Change type

   ACCOUNTABILITY_AND_SOURCE_QUALITY_APPEND_ONLY_NO_WORDING_CHANGE

   Accepted correction

   NO

   Repository adjudication

   NONE

   Wording SHA-256

   cb2fa8aaa0207a90deffb4195a0d359429f28c4c4ab1e2d263c987dd7f4d1a62

   WIP.71 adds attribution-safe intake, external-receipt verifier parity, target-level correction provenance, queue audit-chain controls, human-pilot governance rehearsals, source-quality panels, and currentness v14 without changing claim wording or real authority counts.

   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.

10. WIP72-EVIDENCE-QUARANTINE-AND-DECISION-CONTROL-TRACEABILITY

    Effective date

    2026-08-01

    Change type

    EVIDENCE_GOVERNANCE_AND_SOURCE_TRACEABILITY_NO_WORDING_CHANGE

    Accepted correction

    NO

    Repository adjudication

    NOT_COMPLETED

    Wording SHA-256

    cb2fa8aaa0207a90deffb4195a0d359429f28c4c4ab1e2d263c987dd7f4d1a62

    WIP.72 adds fail-closed evidence quarantine, receipt reconciliation, correction conflict handling, human-pilot preparation, and AI kill-chain citation traceability without changing this claim 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.

• Report of the Royal Commission into the Robodebt Scheme

  Source ID

  AUTH-04-ROBODEBT-ROYAL-COMMISSION

  Class

  OFFICIAL_INQUIRY_REPORT

  Relationship

  PRIMARY_OFFICIAL_INQUIRY

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  Australian Government — Royal Commission into the Robodebt Scheme

  Publication / access

  2023-07-07 / 2026-07-28

  Supports: Official inquiry findings and chronology · Institutional process failures · Published testimony and affected-person evidence within the inquiry record

  Does not support: That every individual harm or death had one exclusive cause · That all automated public-benefit systems share the same design · A universal remedy outcome

• New Robodebt class action settlement

  Source ID

  AUTH-35-ROBODEBT-2026-SETTLEMENT

  Class

  OFFICIAL_SETTLEMENT_IMPLEMENTATION_NOTICE

  Relationship

  OFFICIAL_NOTICE_OF_FEDERAL_COURT_APPROVED_SETTLEMENT

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  Services Australia / Federal Court-approved settlement information

  Publication / access

  2026-06-26 / 2026-07-28

  Supports: Settlement approval date · Additional compensation amount and relation to earlier refunds and debt zeroing · Appointment of a settlement administration process

  Does not support: Individual complete repair · Correction of every downstream record or inference · A merits judgment on every disputed issue

• Income Apportionment Resolution Scheme

  Source ID

  AUTH-36-ROBODEBT-INCOME-APPORTIONMENT-RESOLUTION

  Class

  OFFICIAL_REMEDY_PROGRAM_NOTICE

  Relationship

  PRIMARY_OPERATOR_REMEDY_PROGRAM_DESCRIPTION

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  Services Australia

  Publication / access

  2026-06-11 / 2026-07-28

  Supports: Existence, eligibility pathway, and dates of the resolution scheme · A separate remedy mechanism for eligible affected people

  Does not support: Take-up, accessibility, approval, or rejection rates · Complete compensation · Downstream data correction or repeat-error prevention

• Emotion AI in Job Interviews: Injustice, Emotional Labor, and Design Implications

  Source ID

  AUTH-43-ACM-JOB-INTERVIEW-EMOTION-AI

  Class

  PEER_REVIEWED_USER_STUDY

  Relationship

  PEER_REVIEWED_AFFECTED_USER_STUDY

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  ACM / peer-reviewed HCI research

  Publication / access

  2025-06-23 / 2026-07-28

  Supports: Affected-user perceptions of procedural justice in the studied interview context · Potential emotional labor and fairness burdens

  Does not support: Construct validity of the emotion score · Hiring outcome benefit or predictive validity · Generalization to every hiring system

The first 104 entries preserve the exact WIP.70 baseline. The ninth entry per claim is a WIP.72 attribution-safe intake and source-quality event. No entry is an accepted correction or completed specialist disposition.

1. PRE-WIP58-RETAINED

   Effective date

   BEFORE_2026-07-28

   Change type

   RETAINED_HISTORICAL_WORDING

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   24653f661cd21a1634aa0d9fa21ae0ffc548df7237a16e6b62b48fc7c5822665

   Exact prior wording retained from the versioned correction ledger; this is not an accepted-correction event.

   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.

2. WIP58-CURRENTNESS-REFRESH

   Effective date

   2026-07-28

   Change type

   NO_SUBSTANTIVE_CHANGE

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   d55f61ab535195e1e73271068dc0c61a0ccccd02000ea63ee3054b4edf71e2e8

   Evidence-type and representativeness boundaries remain unchanged.

   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.

3. WIP65-CANONICAL

   Effective date

   2026-07-30

   Change type

   CANONICAL_OPERATION_NORMALIZATION_AND_CURRENTNESS_REFRESH

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   d55f61ab535195e1e73271068dc0c61a0ccccd02000ea63ee3054b4edf71e2e8

   WIP.65 canonicalized the live operation and public status; no accepted correction or real specialist review occurred.

   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.

4. WIP66-INTEROPERABILITY

   Effective date

   2026-07-31

   Change type

   CURRENTNESS_REFRESH_AND_MACHINE_EXPORT_INTEROPERABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   d55f61ab535195e1e73271068dc0c61a0ccccd02000ea63ee3054b4edf71e2e8

   WIP.66 adds deterministic machine export, public version history, queue and retest accountability, and currentness v9 without changing claim 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.

5. WIP67-EXPORT-V2-COMPATIBILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_V2_AND_EXTERNAL_ACKNOWLEDGMENT

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   d55f61ab535195e1e73271068dc0c61a0ccccd02000ea63ee3054b4edf71e2e8

   WIP.67 adds deterministic schema-v2 export, tested schema-v1 compatibility, immutable release-delta receipts, review-queue lifecycle fixtures, correction acknowledgment states, synthetic-only human-pilot dry runs, and currentness version 10 without changing this claim's 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.

6. WIP68-CONFORMANCE-AND-CORRECTION-OBSERVABILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_CONFORMANCE_AND_CORRECTION_OBSERVABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   d55f61ab535195e1e73271068dc0c61a0ccccd02000ea63ee3054b4edf71e2e8

   WIP.68 adds formal schema-v2/v1 conformance receipts, public correction observability, fail-closed queue-recovery fixtures, explicit external request and acknowledgment states, human execution-kit conformance, and currentness version 11 without changing this claim's 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.

7. WIP69-SCHEMA-EVOLUTION-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_SCHEMA_EVOLUTION_AND_EXTERNAL_DELIVERY_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   d55f61ab535195e1e73271068dc0c61a0ccccd02000ea63ee3054b4edf71e2e8

   WIP.69 adds additive schema migration/deprecation governance, prepared external delivery envelopes, queue SLA accountability, correction dependency graphs, and human-pilot authorization gates without changing claim 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.

8. WIP70-CONSUMER-MIGRATION-AND-REHEARSAL-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_CONSUMER_MIGRATION_REHEARSAL_AND_CORRECTION_EXECUTION_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   d55f61ab535195e1e73271068dc0c61a0ccccd02000ea63ee3054b4edf71e2e8

   WIP.70 adds consumer-migration evidence classification, synthetic external-dispatch rehearsal, correction-graph execution simulation, queue-operator accountability, and human-pilot preauthorization without changing claim 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.

9. WIP.71

   Effective date

   2026-08-01

   Change type

   ACCOUNTABILITY_AND_SOURCE_QUALITY_APPEND_ONLY_NO_WORDING_CHANGE

   Accepted correction

   NO

   Repository adjudication

   NONE

   Wording SHA-256

   d55f61ab535195e1e73271068dc0c61a0ccccd02000ea63ee3054b4edf71e2e8

   WIP.71 adds attribution-safe intake, external-receipt verifier parity, target-level correction provenance, queue audit-chain controls, human-pilot governance rehearsals, source-quality panels, and currentness v14 without changing claim wording or real authority counts.

   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.

10. WIP72-EVIDENCE-QUARANTINE-AND-DECISION-CONTROL-TRACEABILITY

    Effective date

    2026-08-01

    Change type

    EVIDENCE_GOVERNANCE_AND_SOURCE_TRACEABILITY_NO_WORDING_CHANGE

    Accepted correction

    NO

    Repository adjudication

    NOT_COMPLETED

    Wording SHA-256

    d55f61ab535195e1e73271068dc0c61a0ccccd02000ea63ee3054b4edf71e2e8

    WIP.72 adds fail-closed evidence quarantine, receipt reconciliation, correction conflict handling, human-pilot preparation, and AI kill-chain citation traceability without changing this claim 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.

• FTC inquiry into AI chatbots acting as companions

  Source ID

  AUTH-01-FTC-COMPANION-INQUIRY

  Class

  OFFICIAL_REGULATORY_INQUIRY

  Relationship

  PRIMARY_ISSUING_BODY_NOTICE

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  U.S. Federal Trade Commission

  Publication / access

  2025-09-11 / 2026-07-28

  Supports: Existence and scope of the FTC inquiry · The companies and issue areas named by the FTC · Regulatory concern requiring further information

  Does not support: A completed enforcement finding · Clinical causation · Product-specific liability or prevalence of harm

• Prepared remarks describing the FTC companion-chatbot 6(b) inquiry

  Source ID

  AUTH-33-FTC-2026-COMPANION-REMARKS

  Class

  OFFICIAL_CURRENTNESS_STATEMENT

  Relationship

  CURRENT_OFFICIAL_STATEMENT_REFERENCING_OPEN_INQUIRY

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  U.S. Federal Trade Commission — Bureau of Consumer Protection

  Publication / access

  2026-03-12 / 2026-07-28

  Supports: The inquiry remained an active information-gathering matter in March 2026 · The seven-company scope and subject areas

  Does not support: Public findings, final enforcement, settlement, or closure · Product-specific wrongdoing · Clinical causality

• Federal Trade Commission oversight testimony describing the companion-chatbot 6(b) study

  Source ID

  AUTH-49-FTC-2026-OVERSIGHT-TESTIMONY

  Class

  OFFICIAL_CONGRESSIONAL_TESTIMONY

  Relationship

  PRIMARY_AGENCY_CURRENTNESS_RECORD

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  U.S. Federal Trade Commission

  Publication / access

  2026-04-15 / 2026-07-28

  Supports: The inquiry remained part of the FTC’s publicly described work in April 2026 · The inquiry is a study rather than a completed enforcement finding · The agency continued to describe the study’s child- and teen-safety scope

  Does not support: A public final report, closure, settlement, or enforcement disposition · Product-specific wrongdoing or clinical causation · The contents of confidential company responses

The first 104 entries preserve the exact WIP.70 baseline. The ninth entry per claim is a WIP.72 attribution-safe intake and source-quality event. No entry is an accepted correction or completed specialist disposition.

1. PRE-WIP58-RETAINED

   Effective date

   BEFORE_2026-07-28

   Change type

   RETAINED_HISTORICAL_WORDING

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   e4c43349c2e0f201beec4751b7204753f6e8ef51fbf6e114a9090ef9dd312152

   Exact prior wording retained from the versioned correction ledger; this is not an accepted-correction event.

   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.

2. WIP58-CURRENTNESS-REFRESH

   Effective date

   2026-07-28

   Change type

   SOURCE_ADDITION

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   1b3e56c698a1acdaa35491f0f6ef35f715176c8bee093dea594ca69b6f5f0209

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

   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.

3. WIP65-CANONICAL

   Effective date

   2026-07-30

   Change type

   CANONICAL_OPERATION_NORMALIZATION_AND_CURRENTNESS_REFRESH

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   1b3e56c698a1acdaa35491f0f6ef35f715176c8bee093dea594ca69b6f5f0209

   WIP.65 canonicalized the live operation and public status; no accepted correction or real specialist review occurred.

   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.

4. WIP66-INTEROPERABILITY

   Effective date

   2026-07-31

   Change type

   CURRENTNESS_REFRESH_AND_MACHINE_EXPORT_INTEROPERABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   1b3e56c698a1acdaa35491f0f6ef35f715176c8bee093dea594ca69b6f5f0209

   WIP.66 adds deterministic machine export, public version history, queue and retest accountability, and currentness v9 without changing claim 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.

5. WIP67-EXPORT-V2-COMPATIBILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_V2_AND_EXTERNAL_ACKNOWLEDGMENT

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   1b3e56c698a1acdaa35491f0f6ef35f715176c8bee093dea594ca69b6f5f0209

   WIP.67 adds deterministic schema-v2 export, tested schema-v1 compatibility, immutable release-delta receipts, review-queue lifecycle fixtures, correction acknowledgment states, synthetic-only human-pilot dry runs, and currentness version 10 without changing this claim's 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.

6. WIP68-CONFORMANCE-AND-CORRECTION-OBSERVABILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_CONFORMANCE_AND_CORRECTION_OBSERVABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   1b3e56c698a1acdaa35491f0f6ef35f715176c8bee093dea594ca69b6f5f0209

   WIP.68 adds formal schema-v2/v1 conformance receipts, public correction observability, fail-closed queue-recovery fixtures, explicit external request and acknowledgment states, human execution-kit conformance, and currentness version 11 without changing this claim's 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.

7. WIP69-SCHEMA-EVOLUTION-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_SCHEMA_EVOLUTION_AND_EXTERNAL_DELIVERY_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   1b3e56c698a1acdaa35491f0f6ef35f715176c8bee093dea594ca69b6f5f0209

   WIP.69 adds additive schema migration/deprecation governance, prepared external delivery envelopes, queue SLA accountability, correction dependency graphs, and human-pilot authorization gates without changing claim 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.

8. WIP70-CONSUMER-MIGRATION-AND-REHEARSAL-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_CONSUMER_MIGRATION_REHEARSAL_AND_CORRECTION_EXECUTION_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   1b3e56c698a1acdaa35491f0f6ef35f715176c8bee093dea594ca69b6f5f0209

   WIP.70 adds consumer-migration evidence classification, synthetic external-dispatch rehearsal, correction-graph execution simulation, queue-operator accountability, and human-pilot preauthorization without changing claim 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.

9. WIP.71

   Effective date

   2026-08-01

   Change type

   ACCOUNTABILITY_AND_SOURCE_QUALITY_APPEND_ONLY_NO_WORDING_CHANGE

   Accepted correction

   NO

   Repository adjudication

   NONE

   Wording SHA-256

   1b3e56c698a1acdaa35491f0f6ef35f715176c8bee093dea594ca69b6f5f0209

   WIP.71 adds attribution-safe intake, external-receipt verifier parity, target-level correction provenance, queue audit-chain controls, human-pilot governance rehearsals, source-quality panels, and currentness v14 without changing claim wording or real authority counts.

   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.

10. WIP72-EVIDENCE-QUARANTINE-AND-DECISION-CONTROL-TRACEABILITY

    Effective date

    2026-08-01

    Change type

    EVIDENCE_GOVERNANCE_AND_SOURCE_TRACEABILITY_NO_WORDING_CHANGE

    Accepted correction

    NO

    Repository adjudication

    NOT_COMPLETED

    Wording SHA-256

    1b3e56c698a1acdaa35491f0f6ef35f715176c8bee093dea594ca69b6f5f0209

    WIP.72 adds fail-closed evidence quarantine, receipt reconciliation, correction conflict handling, human-pilot preparation, and AI kill-chain citation traceability without changing this claim 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.

• FCC final action concerning AI-generated Biden robocalls

  Source ID

  AUTH-02-FCC-BIDEN-ROBOCALL-FINE

  Class

  OFFICIAL_FINAL_ENFORCEMENT_NOTICE

  Relationship

  PRIMARY_ISSUING_BODY_FINAL_ACTION

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  U.S. Federal Communications Commission

  Publication / access

  2024-09-26 / 2026-07-28

  Supports: Artifact and distribution channel existence · Named operator in the FCC action · Spoofing and robocall enforcement disposition

  Does not support: How many recipients believed the call · Any measurable change in voting behavior · Broader strategic election effect

• Disrupting deceptive uses of AI by covert influence operations

  Source ID

  AUTH-19-OPENAI-COVERT-IO

  Class

  PLATFORM_THREAT_INTELLIGENCE_REPORT

  Relationship

  PRIMARY_PROVIDER_REPORT

  Independence

  PARTIAL_OR_UNRESOLVED

  Issuing body

  OpenAI

  Publication / access

  2024-05-30 / 2026-07-28

  Supports: Provider-observed misuse patterns and disruptions · Bounded lack of observed reach/engagement uplift in provider data

  Does not support: No exposure or effect anywhere · No off-platform activity · Independent attribution beyond provider evidence

• FCC report to Congress on robocall enforcement

  Source ID

  AUTH-34-FCC-2025-ROBOCALL-CONGRESS-REPORT

  Class

  OFFICIAL_ENFORCEMENT_REPORT

  Relationship

  PRIMARY_AGENCY_REPORT_TO_CONGRESS

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  Federal Communications Commission

  Publication / access

  2025-12-23 / 2026-07-28

  Supports: Final FCC forfeiture order status · The agency’s 2024 enforcement and collection reporting context

  Does not support: Payment, collection, modification, or judicial disposition after the period covered · Voter behavior or election outcome effects · State-level proceedings not described in the report

• Russian War Report: hacked news program and Zelenskyy surrender deepfake

  Source ID

  AUTH-44-UKRAINE-DFRLAB-ZELENSKYY

  Class

  INDEPENDENT_OPEN_SOURCE_INCIDENT_ANALYSIS

  Relationship

  INDEPENDENT_OSINT_INCIDENT_DOCUMENTATION

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  Digital Forensic Research Lab

  Publication / access

  2022-03-16 / 2026-07-28

  Supports: Artifact existence, distribution channels, and rapid public debunking · Observed ridicule and some contrary interpretations in online comments

  Does not support: Representative recipient exposure or belief · Frontline disruption · Operator identity, sponsorship chain, behavior change, or strategic effect

• FCC Agency Financial Report recording referral of the Steve Kramer forfeiture to DOJ for collection

  Source ID

  AUTH-50-FCC-FY2024-FINANCIAL-REPORT

  Class

  OFFICIAL_AGENCY_FINANCIAL_REPORT

  Relationship

  PRIMARY_AGENCY_POST_ORDER_STATUS_RECORD

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  U.S. Federal Communications Commission

  Publication / access

  2024-11-15 / 2026-07-28

  Supports: Post-order referral to DOJ for collection · Continued agency treatment of the action as an enforcement matter · A distinction between the forfeiture order and actual collection

  Does not support: Payment, collection, settlement, or a judicial judgment collecting the forfeiture · Recipient belief or voting behavior · A final state-law disposition for every related proceeding

The first 104 entries preserve the exact WIP.70 baseline. The ninth entry per claim is a WIP.72 attribution-safe intake and source-quality event. No entry is an accepted correction or completed specialist disposition.

1. PRE-WIP58-RETAINED

   Effective date

   BEFORE_2026-07-28

   Change type

   RETAINED_HISTORICAL_WORDING

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   7fe903cf6f72e7abac78b6221aa01a4ab657b4eddce49b0edffe6ba77fe92ca9

   Exact prior wording retained from the versioned correction ledger; this is not an accepted-correction event.

   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.

2. WIP58-CURRENTNESS-REFRESH

   Effective date

   2026-07-28

   Change type

   NO_SUBSTANTIVE_CHANGE

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cacb10c7d46a7b62e06cb46edf7c9dde9873ed6857acad9a22f96a055dd7253e

   The complete effect chain remains the controlling interpretation.

   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.

3. WIP65-CANONICAL

   Effective date

   2026-07-30

   Change type

   CANONICAL_OPERATION_NORMALIZATION_AND_CURRENTNESS_REFRESH

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cacb10c7d46a7b62e06cb46edf7c9dde9873ed6857acad9a22f96a055dd7253e

   WIP.65 canonicalized the live operation and public status; no accepted correction or real specialist review occurred.

   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.

4. WIP66-INTEROPERABILITY

   Effective date

   2026-07-31

   Change type

   CURRENTNESS_REFRESH_AND_MACHINE_EXPORT_INTEROPERABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cacb10c7d46a7b62e06cb46edf7c9dde9873ed6857acad9a22f96a055dd7253e

   WIP.66 adds deterministic machine export, public version history, queue and retest accountability, and currentness v9 without changing claim 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.

5. WIP67-EXPORT-V2-COMPATIBILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_V2_AND_EXTERNAL_ACKNOWLEDGMENT

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cacb10c7d46a7b62e06cb46edf7c9dde9873ed6857acad9a22f96a055dd7253e

   WIP.67 adds deterministic schema-v2 export, tested schema-v1 compatibility, immutable release-delta receipts, review-queue lifecycle fixtures, correction acknowledgment states, synthetic-only human-pilot dry runs, and currentness version 10 without changing this claim's 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.

6. WIP68-CONFORMANCE-AND-CORRECTION-OBSERVABILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_CONFORMANCE_AND_CORRECTION_OBSERVABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cacb10c7d46a7b62e06cb46edf7c9dde9873ed6857acad9a22f96a055dd7253e

   WIP.68 adds formal schema-v2/v1 conformance receipts, public correction observability, fail-closed queue-recovery fixtures, explicit external request and acknowledgment states, human execution-kit conformance, and currentness version 11 without changing this claim's 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.

7. WIP69-SCHEMA-EVOLUTION-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_SCHEMA_EVOLUTION_AND_EXTERNAL_DELIVERY_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cacb10c7d46a7b62e06cb46edf7c9dde9873ed6857acad9a22f96a055dd7253e

   WIP.69 adds additive schema migration/deprecation governance, prepared external delivery envelopes, queue SLA accountability, correction dependency graphs, and human-pilot authorization gates without changing claim 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.

8. WIP70-CONSUMER-MIGRATION-AND-REHEARSAL-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_CONSUMER_MIGRATION_REHEARSAL_AND_CORRECTION_EXECUTION_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cacb10c7d46a7b62e06cb46edf7c9dde9873ed6857acad9a22f96a055dd7253e

   WIP.70 adds consumer-migration evidence classification, synthetic external-dispatch rehearsal, correction-graph execution simulation, queue-operator accountability, and human-pilot preauthorization without changing claim 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.

9. WIP.71

   Effective date

   2026-08-01

   Change type

   ACCOUNTABILITY_AND_SOURCE_QUALITY_APPEND_ONLY_NO_WORDING_CHANGE

   Accepted correction

   NO

   Repository adjudication

   NONE

   Wording SHA-256

   cacb10c7d46a7b62e06cb46edf7c9dde9873ed6857acad9a22f96a055dd7253e

   WIP.71 adds attribution-safe intake, external-receipt verifier parity, target-level correction provenance, queue audit-chain controls, human-pilot governance rehearsals, source-quality panels, and currentness v14 without changing claim wording or real authority counts.

   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.

10. WIP72-EVIDENCE-QUARANTINE-AND-DECISION-CONTROL-TRACEABILITY

    Effective date

    2026-08-01

    Change type

    EVIDENCE_GOVERNANCE_AND_SOURCE_TRACEABILITY_NO_WORDING_CHANGE

    Accepted correction

    NO

    Repository adjudication

    NOT_COMPLETED

    Wording SHA-256

    cacb10c7d46a7b62e06cb46edf7c9dde9873ed6857acad9a22f96a055dd7253e

    WIP.72 adds fail-closed evidence quarantine, receipt reconciliation, correction conflict handling, human-pilot preparation, and AI kill-chain citation traceability without changing this claim 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.

• Meta 2022 coordinated inauthentic behavior enforcement report

  Source ID

  AUTH-18-META-CIB-2022

  Class

  PLATFORM_THREAT_REPORT

  Relationship

  PRIMARY_PLATFORM_ENFORCEMENT_REPORT

  Independence

  PARTIAL_OR_UNRESOLVED

  Issuing body

  Meta

  Publication / access

  2022-12-15 / 2026-07-28

  Supports: Platform-observed use of synthetic profile imagery · Behavior-centered CIB detection principle

  Does not support: State sponsorship from an image alone · AI authorship of all content · Audience persuasion or strategic effect

• Disrupting deceptive uses of AI by covert influence operations

  Source ID

  AUTH-19-OPENAI-COVERT-IO

  Class

  PLATFORM_THREAT_INTELLIGENCE_REPORT

  Relationship

  PRIMARY_PROVIDER_REPORT

  Independence

  PARTIAL_OR_UNRESOLVED

  Issuing body

  OpenAI

  Publication / access

  2024-05-30 / 2026-07-28

  Supports: Provider-observed misuse patterns and disruptions · Bounded lack of observed reach/engagement uplift in provider data

  Does not support: No exposure or effect anywhere · No off-platform activity · Independent attribution beyond provider evidence

• VIGINUM 2024 activity report

  Source ID

  AUTH-46-VIGINUM-ACTIVITY-2024

  Class

  OFFICIAL_ACTIVITY_AND_GOVERNANCE_REPORT

  Relationship

  PRIMARY_OPERATOR_TRANSPARENCY_REPORT

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  VIGINUM / Secretariat-General for National Defence and Security

  Publication / access

  2025-12-30 / 2026-07-28

  Supports: Published mandate and data-governance controls · Publicly stated minimization, retention, and oversight mechanisms · Operational activity described by the institution

  Does not support: Independent false-positive rate · Incident-level appeals or correction outcomes · Affected-community experience

• Protection of public debate during the March 2026 municipal elections

  Source ID

  AUTH-48-VIGINUM-MUNICIPAL-2026

  Class

  OFFICIAL_ELECTION_PERIOD_REPORT

  Relationship

  PRIMARY_OPERATOR_ELECTION_REPORT

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  VIGINUM / election coordination network

  Publication / access

  2026-06-11 / 2026-07-28

  Supports: Four operations reported by VIGINUM for the election period · Public reporting and threat-characterization workflow · Institutional attribution confidence as stated in the report

  Does not support: Independent verification of every attribution · False-positive, appeal, or correction outcomes · Voter belief or behavior change

The first 104 entries preserve the exact WIP.70 baseline. The ninth entry per claim is a WIP.72 attribution-safe intake and source-quality event. No entry is an accepted correction or completed specialist disposition.

1. PRE-WIP58-RETAINED

   Effective date

   BEFORE_2026-07-28

   Change type

   RETAINED_HISTORICAL_WORDING

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   f4749b3df5315a4f31d23a51d160f335503e6ea87bc00b87c8a3016f23ace8d4

   Exact prior wording retained from the versioned correction ledger; this is not an accepted-correction event.

   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.

2. WIP58-CURRENTNESS-REFRESH

   Effective date

   2026-07-28

   Change type

   CLARIFICATION

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   de23adfd1d91b4cfdee1ae7c5a218092e6fc9f43ad5c2224076dbe99a14514b6

   Separated coordination evidence from synthetic-origin and sponsor attribution.

   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.

3. WIP65-CANONICAL

   Effective date

   2026-07-30

   Change type

   CANONICAL_OPERATION_NORMALIZATION_AND_CURRENTNESS_REFRESH

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   de23adfd1d91b4cfdee1ae7c5a218092e6fc9f43ad5c2224076dbe99a14514b6

   WIP.65 canonicalized the live operation and public status; no accepted correction or real specialist review occurred.

   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.

4. WIP66-INTEROPERABILITY

   Effective date

   2026-07-31

   Change type

   CURRENTNESS_REFRESH_AND_MACHINE_EXPORT_INTEROPERABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   de23adfd1d91b4cfdee1ae7c5a218092e6fc9f43ad5c2224076dbe99a14514b6

   WIP.66 adds deterministic machine export, public version history, queue and retest accountability, and currentness v9 without changing claim 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.

5. WIP67-EXPORT-V2-COMPATIBILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_V2_AND_EXTERNAL_ACKNOWLEDGMENT

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   de23adfd1d91b4cfdee1ae7c5a218092e6fc9f43ad5c2224076dbe99a14514b6

   WIP.67 adds deterministic schema-v2 export, tested schema-v1 compatibility, immutable release-delta receipts, review-queue lifecycle fixtures, correction acknowledgment states, synthetic-only human-pilot dry runs, and currentness version 10 without changing this claim's 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.

6. WIP68-CONFORMANCE-AND-CORRECTION-OBSERVABILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_CONFORMANCE_AND_CORRECTION_OBSERVABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   de23adfd1d91b4cfdee1ae7c5a218092e6fc9f43ad5c2224076dbe99a14514b6

   WIP.68 adds formal schema-v2/v1 conformance receipts, public correction observability, fail-closed queue-recovery fixtures, explicit external request and acknowledgment states, human execution-kit conformance, and currentness version 11 without changing this claim's 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.

7. WIP69-SCHEMA-EVOLUTION-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_SCHEMA_EVOLUTION_AND_EXTERNAL_DELIVERY_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   de23adfd1d91b4cfdee1ae7c5a218092e6fc9f43ad5c2224076dbe99a14514b6

   WIP.69 adds additive schema migration/deprecation governance, prepared external delivery envelopes, queue SLA accountability, correction dependency graphs, and human-pilot authorization gates without changing claim 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.

8. WIP70-CONSUMER-MIGRATION-AND-REHEARSAL-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_CONSUMER_MIGRATION_REHEARSAL_AND_CORRECTION_EXECUTION_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   de23adfd1d91b4cfdee1ae7c5a218092e6fc9f43ad5c2224076dbe99a14514b6

   WIP.70 adds consumer-migration evidence classification, synthetic external-dispatch rehearsal, correction-graph execution simulation, queue-operator accountability, and human-pilot preauthorization without changing claim 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.

9. WIP.71

   Effective date

   2026-08-01

   Change type

   ACCOUNTABILITY_AND_SOURCE_QUALITY_APPEND_ONLY_NO_WORDING_CHANGE

   Accepted correction

   NO

   Repository adjudication

   NONE

   Wording SHA-256

   de23adfd1d91b4cfdee1ae7c5a218092e6fc9f43ad5c2224076dbe99a14514b6

   WIP.71 adds attribution-safe intake, external-receipt verifier parity, target-level correction provenance, queue audit-chain controls, human-pilot governance rehearsals, source-quality panels, and currentness v14 without changing claim wording or real authority counts.

   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.

10. WIP72-EVIDENCE-QUARANTINE-AND-DECISION-CONTROL-TRACEABILITY

    Effective date

    2026-08-01

    Change type

    EVIDENCE_GOVERNANCE_AND_SOURCE_TRACEABILITY_NO_WORDING_CHANGE

    Accepted correction

    NO

    Repository adjudication

    NOT_COMPLETED

    Wording SHA-256

    de23adfd1d91b4cfdee1ae7c5a218092e6fc9f43ad5c2224076dbe99a14514b6

    WIP.72 adds fail-closed evidence quarantine, receipt reconciliation, correction conflict handling, human-pilot preparation, and AI kill-chain citation traceability without changing this claim 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.

• Analytical report on AI-generated videos used to discredit Ukraine’s Defence Forces

  Source ID

  AUTH-22-UKRAINE-CPD-DEEPFAKES

  Class

  OFFICIAL_INCIDENT_REPORT

  Relationship

  PRIMARY_TARGETED_STATE_REPORT

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  Center for Countering Disinformation, National Security and Defense Council of Ukraine

  Publication / access

  2024-02-01 / 2026-07-28

  Supports: Artifact existence and official response · Incident chronology and targeted institution’s assessment

  Does not support: Precise operator identity for every artifact · Continuing audience belief · Measured operational or strategic effect

• C2PA Technical Specification 2.4

  Source ID

  AUTH-07-C2PA-SPEC-2-4

  Class

  TECHNICAL_STANDARD

  Relationship

  PRIMARY_STANDARD_TEXT

  Independence

  PARTIAL_OR_UNRESOLVED

  Issuing body

  Coalition for Content Provenance and Authenticity

  Publication / access

  2025-11-01 / 2026-07-28

  Supports: How C2PA provenance records are structured · Cryptographic verification of signed provenance claims

  Does not support: Truth of the depicted event · That unsigned media is fake · That credentials survive every transformation or platform

• Strengthening Multimedia Integrity in the Generative AI Era

  Source ID

  AUTH-38-C2PA-NSA-GUIDANCE

  Class

  OFFICIAL_CYBERSECURITY_GUIDANCE

  Relationship

  MULTI_AGENCY_TECHNICAL_GUIDANCE

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  NSA, CISA, FBI, and international cybersecurity partners

  Publication / access

  2025-01-16 / 2026-07-28

  Supports: Layered provenance architecture and operational deployment guidance · The need to combine metadata with watermarking and fingerprinting to improve resilience · Provenance supports claims about origin and history, not truth of depicted events

  Does not support: Independent real-world survival rates across all platforms · Universal adoption · Proof that unsigned media is false

• Russian War Report: hacked news program and Zelenskyy surrender deepfake

  Source ID

  AUTH-44-UKRAINE-DFRLAB-ZELENSKYY

  Class

  INDEPENDENT_OPEN_SOURCE_INCIDENT_ANALYSIS

  Relationship

  INDEPENDENT_OSINT_INCIDENT_DOCUMENTATION

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  Digital Forensic Research Lab

  Publication / access

  2022-03-16 / 2026-07-28

  Supports: Artifact existence, distribution channels, and rapid public debunking · Observed ridicule and some contrary interpretations in online comments

  Does not support: Representative recipient exposure or belief · Frontline disruption · Operator identity, sponsorship chain, behavior change, or strategic effect

• Do deepfake videos undermine our epistemic trust?

  Source ID

  AUTH-45-UKRAINE-PLOS-DEEPFAKE-TRUST

  Class

  PEER_REVIEWED_SOCIAL_MEDIA_DISCOURSE_STUDY

  Relationship

  PEER_REVIEWED_THEMATIC_ANALYSIS

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  PLOS ONE

  Publication / access

  2023-10-25 / 2026-07-28

  Supports: Evidence of online discourse and epistemic skepticism · Practical examples of liar’s-dividend-like claims in a social-media sample

  Does not support: Representative population belief · Causal behavior change · Frontline operational effect

The first 104 entries preserve the exact WIP.70 baseline. The ninth entry per claim is a WIP.72 attribution-safe intake and source-quality event. No entry is an accepted correction or completed specialist disposition.

1. PRE-WIP58-RETAINED

   Effective date

   BEFORE_2026-07-28

   Change type

   RETAINED_HISTORICAL_WORDING

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   56a0695e448e5dcfd769b141a45614249998d1368711f215263b09ac630d25a7

   Exact prior wording retained from the versioned correction ledger; this is not an accepted-correction event.

   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.

2. WIP58-CURRENTNESS-REFRESH

   Effective date

   2026-07-28

   Change type

   NARROWING

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   569cca2f52b0142fd96622e11cb125bb20feb5271bd46b80b1861108bf1bc6c2

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

   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.

3. WIP65-CANONICAL

   Effective date

   2026-07-30

   Change type

   CANONICAL_OPERATION_NORMALIZATION_AND_CURRENTNESS_REFRESH

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   569cca2f52b0142fd96622e11cb125bb20feb5271bd46b80b1861108bf1bc6c2

   WIP.65 canonicalized the live operation and public status; no accepted correction or real specialist review occurred.

   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.

4. WIP66-INTEROPERABILITY

   Effective date

   2026-07-31

   Change type

   CURRENTNESS_REFRESH_AND_MACHINE_EXPORT_INTEROPERABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   569cca2f52b0142fd96622e11cb125bb20feb5271bd46b80b1861108bf1bc6c2

   WIP.66 adds deterministic machine export, public version history, queue and retest accountability, and currentness v9 without changing claim 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.

5. WIP67-EXPORT-V2-COMPATIBILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_V2_AND_EXTERNAL_ACKNOWLEDGMENT

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   569cca2f52b0142fd96622e11cb125bb20feb5271bd46b80b1861108bf1bc6c2

   WIP.67 adds deterministic schema-v2 export, tested schema-v1 compatibility, immutable release-delta receipts, review-queue lifecycle fixtures, correction acknowledgment states, synthetic-only human-pilot dry runs, and currentness version 10 without changing this claim's 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.

6. WIP68-CONFORMANCE-AND-CORRECTION-OBSERVABILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_CONFORMANCE_AND_CORRECTION_OBSERVABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   569cca2f52b0142fd96622e11cb125bb20feb5271bd46b80b1861108bf1bc6c2

   WIP.68 adds formal schema-v2/v1 conformance receipts, public correction observability, fail-closed queue-recovery fixtures, explicit external request and acknowledgment states, human execution-kit conformance, and currentness version 11 without changing this claim's 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.

7. WIP69-SCHEMA-EVOLUTION-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_SCHEMA_EVOLUTION_AND_EXTERNAL_DELIVERY_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   569cca2f52b0142fd96622e11cb125bb20feb5271bd46b80b1861108bf1bc6c2

   WIP.69 adds additive schema migration/deprecation governance, prepared external delivery envelopes, queue SLA accountability, correction dependency graphs, and human-pilot authorization gates without changing claim 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.

8. WIP70-CONSUMER-MIGRATION-AND-REHEARSAL-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_CONSUMER_MIGRATION_REHEARSAL_AND_CORRECTION_EXECUTION_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   569cca2f52b0142fd96622e11cb125bb20feb5271bd46b80b1861108bf1bc6c2

   WIP.70 adds consumer-migration evidence classification, synthetic external-dispatch rehearsal, correction-graph execution simulation, queue-operator accountability, and human-pilot preauthorization without changing claim 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.

9. WIP.71

   Effective date

   2026-08-01

   Change type

   ACCOUNTABILITY_AND_SOURCE_QUALITY_APPEND_ONLY_NO_WORDING_CHANGE

   Accepted correction

   NO

   Repository adjudication

   NONE

   Wording SHA-256

   569cca2f52b0142fd96622e11cb125bb20feb5271bd46b80b1861108bf1bc6c2

   WIP.71 adds attribution-safe intake, external-receipt verifier parity, target-level correction provenance, queue audit-chain controls, human-pilot governance rehearsals, source-quality panels, and currentness v14 without changing claim wording or real authority counts.

   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.

10. WIP72-EVIDENCE-QUARANTINE-AND-DECISION-CONTROL-TRACEABILITY

    Effective date

    2026-08-01

    Change type

    EVIDENCE_GOVERNANCE_AND_SOURCE_TRACEABILITY_NO_WORDING_CHANGE

    Accepted correction

    NO

    Repository adjudication

    NOT_COMPLETED

    Wording SHA-256

    569cca2f52b0142fd96622e11cb125bb20feb5271bd46b80b1861108bf1bc6c2

    WIP.72 adds fail-closed evidence quarantine, receipt reconciliation, correction conflict handling, human-pilot preparation, and AI kill-chain citation traceability without changing this claim 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.

• C2PA Technical Specification 2.4

  Source ID

  AUTH-07-C2PA-SPEC-2-4

  Class

  TECHNICAL_STANDARD

  Relationship

  PRIMARY_STANDARD_TEXT

  Independence

  PARTIAL_OR_UNRESOLVED

  Issuing body

  Coalition for Content Provenance and Authenticity

  Publication / access

  2025-11-01 / 2026-07-28

  Supports: How C2PA provenance records are structured · Cryptographic verification of signed provenance claims

  Does not support: Truth of the depicted event · That unsigned media is fake · That credentials survive every transformation or platform

• The Liar’s Dividend: Can Politicians Claim Misinformation to Evade Accountability?

  Source ID

  AUTH-20-LIARS-DIVIDEND-APSR

  Class

  PEER_REVIEWED_EXPERIMENTAL_RESEARCH

  Relationship

  PRIMARY_PEER_REVIEWED_EXPERIMENTS

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  American Political Science Review authors

  Publication / access

  2023-10-10 / 2026-07-28

  Supports: The liar’s dividend as a measurable but format- and context-dependent phenomenon · Experimental evidence against treating denial as uniformly effective

  Does not support: Effect in every country or crisis · Long-term behavior change · Effectiveness against every media format

• Strengthening Multimedia Integrity in the Generative AI Era

  Source ID

  AUTH-38-C2PA-NSA-GUIDANCE

  Class

  OFFICIAL_CYBERSECURITY_GUIDANCE

  Relationship

  MULTI_AGENCY_TECHNICAL_GUIDANCE

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  NSA, CISA, FBI, and international cybersecurity partners

  Publication / access

  2025-01-16 / 2026-07-28

  Supports: Layered provenance architecture and operational deployment guidance · The need to combine metadata with watermarking and fingerprinting to improve resilience · Provenance supports claims about origin and history, not truth of depicted events

  Does not support: Independent real-world survival rates across all platforms · Universal adoption · Proof that unsigned media is false

• Verifying Provenance of Digital Media: Why the C2PA Security Model Is Not Yet Ready for High-Stakes Use

  Source ID

  AUTH-39-C2PA-INDEPENDENT-FORMAL-ANALYSIS

  Class

  INDEPENDENT_FORMAL_SECURITY_ANALYSIS

  Relationship

  INDEPENDENT_FORMAL_METHODS_RESEARCH

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  Independent academic researchers

  Publication / access

  2026-04-23 / 2026-07-28

  Supports: Documented independent criticism and identified protocol-level risks · The need for high-stakes caution and complementary verification

  Does not support: That every C2PA implementation fails · Field prevalence of each vulnerability · A final standards-body disposition

• Do deepfake videos undermine our epistemic trust?

  Source ID

  AUTH-45-UKRAINE-PLOS-DEEPFAKE-TRUST

  Class

  PEER_REVIEWED_SOCIAL_MEDIA_DISCOURSE_STUDY

  Relationship

  PEER_REVIEWED_THEMATIC_ANALYSIS

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  PLOS ONE

  Publication / access

  2023-10-25 / 2026-07-28

  Supports: Evidence of online discourse and epistemic skepticism · Practical examples of liar’s-dividend-like claims in a social-media sample

  Does not support: Representative population belief · Causal behavior change · Frontline operational effect

• C2PA 2.4 explainer and durable Content Credentials guidance

  Source ID

  AUTH-51-C2PA-2-4-EXPLAINER

  Class

  PRIMARY_TECHNICAL_STANDARD_EXPLAINER

  Relationship

  PRIMARY_STANDARD_SCOPE_AND_LIMITATION_RECORD

  Independence

  PARTIAL_OR_UNRESOLVED

  Issuing body

  Coalition for Content Provenance and Authenticity

  Publication / access

  Version 2.4; accessed 2026-07-28 / 2026-07-28

  Supports: Embedded provenance metadata can be removed · Soft bindings can support discovery of remotely stored credentials after separation · Durability is an implementation architecture rather than a guarantee that every transformation preserves recoverability

  Does not support: 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

• C2PA Technical Specification 2.4 and implementation guidance

  Source ID

  AUTH-38-C2PA-SPEC-2-4

  Class

  PRIMARY_TECHNICAL_STANDARD

  Relationship

  PRIMARY_STANDARD

  Independence

  NOT_RECORDED

  Issuing body

  Coalition for Content Provenance and Authenticity

  Publication / access

  NOT_RECORDED / NOT_RECORDED

  Supports: Defines durable Content Credentials, soft bindings, manifest repositories, and C2PA trust lists.

  Does not support: Truth of depicted events, universal adoption, or reliable credential survival across every crisis chain.

The first 104 entries preserve the exact WIP.70 baseline. The ninth entry per claim is a WIP.72 attribution-safe intake and source-quality event. No entry is an accepted correction or completed specialist disposition.

1. PRE-WIP58-RETAINED

   Effective date

   BEFORE_2026-07-28

   Change type

   RETAINED_HISTORICAL_WORDING

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   64b9e33bf80d5e877e1bb5973307abc5777b5158cace7564eb3e07ded3d23d79

   Exact prior wording retained from the versioned correction ledger; this is not an accepted-correction event.

   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.

2. WIP58-CURRENTNESS-REFRESH

   Effective date

   2026-07-28

   Change type

   CLARIFICATION

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   53a4e3bd5691248658e03951a56449f24c5c533e0444e43a906690e3356126ae

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

   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.

3. WIP65-CANONICAL

   Effective date

   2026-07-30

   Change type

   CANONICAL_OPERATION_NORMALIZATION_AND_CURRENTNESS_REFRESH

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   2d422bad94e7ddf4c4604b8bf5bb0090dd446ce95510b76d0c5e0b1221f2a022

   WIP.65 canonicalized the live operation and public status; no accepted correction or real specialist review occurred.

   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.

4. WIP66-INTEROPERABILITY

   Effective date

   2026-07-31

   Change type

   CURRENTNESS_REFRESH_AND_MACHINE_EXPORT_INTEROPERABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   2d422bad94e7ddf4c4604b8bf5bb0090dd446ce95510b76d0c5e0b1221f2a022

   WIP.66 adds deterministic machine export, public version history, queue and retest accountability, and currentness v9 without changing claim wording.

   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.

5. WIP67-EXPORT-V2-COMPATIBILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_V2_AND_EXTERNAL_ACKNOWLEDGMENT

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   2d422bad94e7ddf4c4604b8bf5bb0090dd446ce95510b76d0c5e0b1221f2a022

   WIP.67 adds deterministic schema-v2 export, tested schema-v1 compatibility, immutable release-delta receipts, review-queue lifecycle fixtures, correction acknowledgment states, synthetic-only human-pilot dry runs, and currentness version 10 without changing this claim's wording.

   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.

6. WIP68-CONFORMANCE-AND-CORRECTION-OBSERVABILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_CONFORMANCE_AND_CORRECTION_OBSERVABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   2d422bad94e7ddf4c4604b8bf5bb0090dd446ce95510b76d0c5e0b1221f2a022

   WIP.68 adds formal schema-v2/v1 conformance receipts, public correction observability, fail-closed queue-recovery fixtures, explicit external request and acknowledgment states, human execution-kit conformance, and currentness version 11 without changing this claim's wording.

   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.

7. WIP69-SCHEMA-EVOLUTION-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_SCHEMA_EVOLUTION_AND_EXTERNAL_DELIVERY_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   2d422bad94e7ddf4c4604b8bf5bb0090dd446ce95510b76d0c5e0b1221f2a022

   WIP.69 adds additive schema migration/deprecation governance, prepared external delivery envelopes, queue SLA accountability, correction dependency graphs, and human-pilot authorization gates without changing claim wording.

   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.

8. WIP70-CONSUMER-MIGRATION-AND-REHEARSAL-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_CONSUMER_MIGRATION_REHEARSAL_AND_CORRECTION_EXECUTION_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   2d422bad94e7ddf4c4604b8bf5bb0090dd446ce95510b76d0c5e0b1221f2a022

   WIP.70 adds consumer-migration evidence classification, synthetic external-dispatch rehearsal, correction-graph execution simulation, queue-operator accountability, and human-pilot preauthorization without changing claim wording.

   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.

9. WIP.71

   Effective date

   2026-08-01

   Change type

   ACCOUNTABILITY_AND_SOURCE_QUALITY_APPEND_ONLY_NO_WORDING_CHANGE

   Accepted correction

   NO

   Repository adjudication

   NONE

   Wording SHA-256

   2d422bad94e7ddf4c4604b8bf5bb0090dd446ce95510b76d0c5e0b1221f2a022

   WIP.71 adds attribution-safe intake, external-receipt verifier parity, target-level correction provenance, queue audit-chain controls, human-pilot governance rehearsals, source-quality panels, and currentness v14 without changing claim wording or real authority counts.

   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.

10. WIP72-EVIDENCE-QUARANTINE-AND-DECISION-CONTROL-TRACEABILITY

    Effective date

    2026-08-01

    Change type

    EVIDENCE_GOVERNANCE_AND_SOURCE_TRACEABILITY_NO_WORDING_CHANGE

    Accepted correction

    NO

    Repository adjudication

    NOT_COMPLETED

    Wording SHA-256

    2d422bad94e7ddf4c4604b8bf5bb0090dd446ce95510b76d0c5e0b1221f2a022

    WIP.72 adds fail-closed evidence quarantine, receipt reconciliation, correction conflict handling, human-pilot preparation, and AI kill-chain citation traceability without changing this claim wording.

    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.

• Information threat linked to artificial intelligence

  Source ID

  AUTH-21-VIGINUM-AI-THREAT

  Class

  OFFICIAL_THREAT_ASSESSMENT

  Relationship

  PRIMARY_ISSUING_BODY_REPORT

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  VIGINUM / French Secretariat-General for National Defence and Security

  Publication / access

  2025-02-07 / 2026-07-28

  Supports: The stated French institutional mandate · Behavior- and foreign-interference-focused defensive framing

  Does not support: That every classification is correct · Authority over lawful domestic belief · Strategic effect of every detected campaign

• VIGINUM 2024 activity report

  Source ID

  AUTH-46-VIGINUM-ACTIVITY-2024

  Class

  OFFICIAL_ACTIVITY_AND_GOVERNANCE_REPORT

  Relationship

  PRIMARY_OPERATOR_TRANSPARENCY_REPORT

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  VIGINUM / Secretariat-General for National Defence and Security

  Publication / access

  2025-12-30 / 2026-07-28

  Supports: Published mandate and data-governance controls · Publicly stated minimization, retention, and oversight mechanisms · Operational activity described by the institution

  Does not support: Independent false-positive rate · Incident-level appeals or correction outcomes · Affected-community experience

• Opinion of the VIGINUM Ethics and Scientific Committee on 2024 activity

  Source ID

  AUTH-47-VIGINUM-ETHICS-2024-OPINION

  Class

  INDEPENDENT_OR_SEPARATE_OVERSIGHT_OPINION

  Relationship

  PUBLIC_OVERSIGHT_COMMITTEE_OPINION

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  VIGINUM Ethics and Scientific Committee

  Publication / access

  2025-12-22 / 2026-07-28

  Supports: Existence and public role of the oversight committee · Committee access to collection information and analysis notes · Public recommendations concerning the legal and institutional framework

  Does not support: A judicially independent appeal forum · Measured false-positive burden · Correction outcomes for specific people or organizations

• Protection of public debate during the March 2026 municipal elections

  Source ID

  AUTH-48-VIGINUM-MUNICIPAL-2026

  Class

  OFFICIAL_ELECTION_PERIOD_REPORT

  Relationship

  PRIMARY_OPERATOR_ELECTION_REPORT

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  VIGINUM / election coordination network

  Publication / access

  2026-06-11 / 2026-07-28

  Supports: Four operations reported by VIGINUM for the election period · Public reporting and threat-characterization workflow · Institutional attribution confidence as stated in the report

  Does not support: Independent verification of every attribution · False-positive, appeal, or correction outcomes · Voter belief or behavior change

• VIGINUM mission page recording the 2026 decree that strengthened the service’s remit

  Source ID

  AUTH-53-VIGINUM-2026-MISSION-DECREE

  Class

  OFFICIAL_MISSION_AND_LEGAL_CURRENTNESS_PAGE

  Relationship

  PRIMARY_OPERATOR_AND_LEGAL_CURRENTNESS_RECORD

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  Secrétariat général de la défense et de la sécurité nationale (SGDSN)

  Publication / access

  2026-02-11 decree; page accessed 2026-07-28 / 2026-07-28

  Supports: The service remained active and its remit was formally updated in 2026 · The official characterization of the mission as defensive · Current operator and public-debate protection scope

  Does not support: Independent validation of attribution decisions · A public false-positive rate or complete correction and appeal system · Affected-community acceptance or behavior effects

• Untitled source

  Source ID

  AUTH-25-VIGINUM-MANDATE

  Class

  UNCLASSIFIED

  Relationship

  UNCLASSIFIED

  Independence

  NOT_RECORDED

  Issuing body

  NOT_RECORDED

  Publication / access

  NOT_RECORDED / NOT_RECORDED

  Supports: NOT_RECORDED

  Does not support: NOT_RECORDED

• CNIL opinion on VIGINUM personal-data processing

  Source ID

  AUTH-41-VIGINUM-CNIL-REVIEW

  Class

  PRIMARY_REGULATORY_OPINION

  Relationship

  PRIMARY_REGULATORY_RECORD

  Independence

  NOT_RECORDED

  Issuing body

  Commission nationale de l'informatique et des libertés / Legifrance

  Publication / access

  NOT_RECORDED / NOT_RECORDED

  Supports: Documents data-minimization concerns, retention limits, oversight, and indirect access, rectification, and erasure routes.

  Does not support: A public register of false positives, successful remedies, affected-community outcomes, or error rates.

The first 104 entries preserve the exact WIP.70 baseline. The ninth entry per claim is a WIP.72 attribution-safe intake and source-quality event. No entry is an accepted correction or completed specialist disposition.

1. PRE-WIP58-RETAINED

   Effective date

   BEFORE_2026-07-28

   Change type

   RETAINED_HISTORICAL_WORDING

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   8d3ff1c0768dea821a63f31989795872fbfee2e80b88f684c51702f228799c36

   Exact prior wording retained from the versioned correction ledger; this is not an accepted-correction event.

   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.

2. WIP58-CURRENTNESS-REFRESH

   Effective date

   2026-07-28

   Change type

   CURRENTNESS_REFRESH

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   c50a0cb3f301ce6bc287dca2557a9332ef1e81e0be1942e7851a77f5516b5fdd

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

   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.

3. WIP65-CANONICAL

   Effective date

   2026-07-30

   Change type

   CANONICAL_OPERATION_NORMALIZATION_AND_CURRENTNESS_REFRESH

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cccec9f9d5e2a028d68e8ef326be53b858974845dbb3469ee942475bbb4bcaeb

   WIP.65 canonicalized the live operation and public status; no accepted correction or real specialist review occurred.

   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.

4. WIP66-INTEROPERABILITY

   Effective date

   2026-07-31

   Change type

   CURRENTNESS_REFRESH_AND_MACHINE_EXPORT_INTEROPERABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cccec9f9d5e2a028d68e8ef326be53b858974845dbb3469ee942475bbb4bcaeb

   WIP.66 adds deterministic machine export, public version history, queue and retest accountability, and currentness v9 without changing claim wording.

   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.

5. WIP67-EXPORT-V2-COMPATIBILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_V2_AND_EXTERNAL_ACKNOWLEDGMENT

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cccec9f9d5e2a028d68e8ef326be53b858974845dbb3469ee942475bbb4bcaeb

   WIP.67 adds deterministic schema-v2 export, tested schema-v1 compatibility, immutable release-delta receipts, review-queue lifecycle fixtures, correction acknowledgment states, synthetic-only human-pilot dry runs, and currentness version 10 without changing this claim's wording.

   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.

6. WIP68-CONFORMANCE-AND-CORRECTION-OBSERVABILITY

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_EXPORT_CONFORMANCE_AND_CORRECTION_OBSERVABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cccec9f9d5e2a028d68e8ef326be53b858974845dbb3469ee942475bbb4bcaeb

   WIP.68 adds formal schema-v2/v1 conformance receipts, public correction observability, fail-closed queue-recovery fixtures, explicit external request and acknowledgment states, human execution-kit conformance, and currentness version 11 without changing this claim's wording.

   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.

7. WIP69-SCHEMA-EVOLUTION-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_SCHEMA_EVOLUTION_AND_EXTERNAL_DELIVERY_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cccec9f9d5e2a028d68e8ef326be53b858974845dbb3469ee942475bbb4bcaeb

   WIP.69 adds additive schema migration/deprecation governance, prepared external delivery envelopes, queue SLA accountability, correction dependency graphs, and human-pilot authorization gates without changing claim wording.

   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.

8. WIP70-CONSUMER-MIGRATION-AND-REHEARSAL-1

   Effective date

   2026-07-31

   Change type

   NO_WORDING_CHANGE_CONSUMER_MIGRATION_REHEARSAL_AND_CORRECTION_EXECUTION_ACCOUNTABILITY

   Accepted correction

   NO

   Repository adjudication

   NOT_COMPLETED

   Wording SHA-256

   cccec9f9d5e2a028d68e8ef326be53b858974845dbb3469ee942475bbb4bcaeb

   WIP.70 adds consumer-migration evidence classification, synthetic external-dispatch rehearsal, correction-graph execution simulation, queue-operator accountability, and human-pilot preauthorization without changing claim wording.

   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.

9. WIP.71

   Effective date

   2026-08-01

   Change type

   ACCOUNTABILITY_AND_SOURCE_QUALITY_APPEND_ONLY_NO_WORDING_CHANGE

   Accepted correction

   NO

   Repository adjudication

   NONE

   Wording SHA-256

   cccec9f9d5e2a028d68e8ef326be53b858974845dbb3469ee942475bbb4bcaeb

   WIP.71 adds attribution-safe intake, external-receipt verifier parity, target-level correction provenance, queue audit-chain controls, human-pilot governance rehearsals, source-quality panels, and currentness v14 without changing claim wording or real authority counts.

   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.

10. WIP72-EVIDENCE-QUARANTINE-AND-DECISION-CONTROL-TRACEABILITY

    Effective date

    2026-08-01

    Change type

    EVIDENCE_GOVERNANCE_AND_SOURCE_TRACEABILITY_NO_WORDING_CHANGE

    Accepted correction

    NO

    Repository adjudication

    NOT_COMPLETED

    Wording SHA-256

    cccec9f9d5e2a028d68e8ef326be53b858974845dbb3469ee942475bbb4bcaeb

    WIP.72 adds fail-closed evidence quarantine, receipt reconciliation, correction conflict handling, human-pilot preparation, and AI kill-chain citation traceability without changing this claim wording.

    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.

• Advisory Concerning the Chicago Police Department’s Predictive Risk Models

  Source ID

  AUTH-11-CHICAGO-PREDICTIVE-RISK

  Class

  OFFICIAL_OVERSIGHT_ADVISORY

  Relationship

  PRIMARY_OVERSIGHT_RECORD

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  City of Chicago Office of Inspector General

  Publication / access

  2020-01-30 / 2026-07-28

  Supports: Existence, function, governance history, and decommissioning · Person-level scoring as distinct from aggregate forecasting

  Does not support: That every predictive-policing system uses the same model · A single causal effect on crime or public trust · Complete downstream deletion of historical scores

• Allegheny Family Screening Tool

  Source ID

  AUTH-12-ALLEGHENY-FAMILY-SCREENING

  Class

  OFFICIAL_SYSTEM_DESCRIPTION

  Relationship

  PRIMARY_OPERATOR_DESCRIPTION

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  Allegheny County Department of Human Services

  Publication / access

  2025-01-01 / 2026-07-28

  Supports: Officially described purpose and decision role · A documented person/family-level risk-screening deployment

  Does not support: Independent accuracy, fairness, or causal benefit · That human judgment neutralizes automation bias · Complete remedy for errors

• District Court of The Hague judgment concerning SyRI

  Source ID

  AUTH-13-SYRI-JUDGMENT

  Class

  JUDICIAL_DECISION

  Relationship

  PRIMARY_JUDICIAL_RECORD

  Independence

  PARTIAL_OR_UNRESOLVED

  Issuing body

  Rechtbank Den Haag (District Court of The Hague)

  Publication / access

  2020-02-05 / 2026-07-28

  Supports: Existence and legal disposition of the SyRI case · Judicial concern about transparency, proportionality, and privacy

  Does not support: That all fraud-risk analytics are unlawful · A universal rule outside the judgment’s jurisdiction and facts · Complete remediation for affected communities

• CDC FluSight forecasting

  Source ID

  AUTH-14-CDC-FLUSIGHT

  Class

  OFFICIAL_PUBLIC_HEALTH_FORECASTING_PROGRAM

  Relationship

  PRIMARY_PROGRAM_DESCRIPTION

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  U.S. Centers for Disease Control and Prevention

  Publication / access

  2026-07-01 / 2026-07-28

  Supports: A beneficial aggregate forecasting use case · Operational distinction between population-level planning and person-level scoring

  Does not support: Perfect forecast accuracy · Authority to take coercive action against individuals · Transferability to political or security prediction

• Home Office Complexity Application Routing Solution — Visits

  Source ID

  AUTH-15-HOME-OFFICE-CARS

  Class

  OFFICIAL_ALGORITHMIC_TRANSPARENCY_RECORD

  Relationship

  PRIMARY_OPERATOR_TRANSPARENCY_RECORD

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  UK Home Office

  Publication / access

  2024-12-17 / 2026-07-28

  Supports: A documented workflow-routing deployment · Distinction between prioritization/routing and final decision authority

  Does not support: Absence of automation bias · Fairness or accuracy across groups · No effect on processing time or applicant burden

• UNHCR predictive analytics for forced displacement

  Source ID

  AUTH-16-UNHCR-DISPLACEMENT-FORECAST

  Class

  OFFICIAL_TECHNICAL_WORKSHOP_RECORD

  Relationship

  PRIMARY_INSTITUTIONAL_TECHNICAL_RECORD

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  United Nations High Commissioner for Refugees

  Publication / access

  2025-04-01 / 2026-07-28

  Supports: A humanitarian aggregate forecasting use case · Planning-oriented rather than person-level punitive application

  Does not support: Reliable prediction of individual movement · Automatic entitlement or enforcement decisions · Universal model performance across crises

• Dynamic Conflict Risk Model technical report

  Source ID

  AUTH-17-EU-JRC-DYNAMIC-CONFLICT-RISK

  Class

  OFFICIAL_RESEARCH_REPORT

  Relationship

  PRIMARY_OFFICIAL_RESEARCH

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  European Commission Joint Research Centre

  Publication / access

  2025-06-01 / 2026-07-28

  Supports: A documented aggregate conflict-forecasting system · Explicit integration of quantitative and qualitative judgment

  Does not support: Certainty about conflict onset · A lawful basis for person-level intervention · Strategic effect of any action taken from a forecast

• Predictions Put Into Practice: A Quasi-Experimental Evaluation of Chicago’s Predictive Policing Pilot

  Source ID

  AUTH-23-CHICAGO-NIJ-EVALUATION

  Class

  OFFICIAL_RESEARCH_SUMMARY

  Relationship

  OFFICIAL_RESEARCH_SUMMARY_OF_QUASI_EXPERIMENTAL_EVALUATION

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  U.S. National Institute of Justice

  Publication / access

  2017-09-01 / 2026-07-28

  Supports: The evaluated SSL pilot involved 426 people estimated to be at highest risk of gun violence · The study used time-series and matched-comparison methods · The evaluation did not find a significant reduction in homicide or shooting victimization attributable to list placement and reported increased likelihood of arrest for a shooting among listed people

  Does not support: Complete individual-level notice or remedy · That every later CPD risk model had the same implementation · Causal proof concerning all predictive-policing systems

• Parliamentary answers concerning SyRI data destruction and decision use

  Source ID

  AUTH-24-SYRI-DATA-DESTRUCTION

  Class

  OFFICIAL_PARLIAMENTARY_ANSWER

  Relationship

  PRIMARY_EXECUTIVE_ANSWER_TO_PARLIAMENT

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  Government of the Netherlands — House of Representatives records

  Publication / access

  2022-06-21 / 2026-07-28

  Supports: An official statement that SyRI-processed data had been destroyed by June 2022 · An official statement that responsible bodies did not make decisions based on SyRI-derived data · SyRI algorithms and criteria were not used in the named Detectie aan de Poort systems

  Does not support: Independent forensic verification of destruction · Deletion from every backup, derivative, vendor, or institutional system · Complete repair for communities subjected to SyRI projects

• Impact Evaluation of the Allegheny Family Screening Tool — Phase 2

  Source ID

  AUTH-25-AFST-PHASE2-EVALUATION

  Class

  OPERATOR_COMMISSIONED_IMPACT_EVALUATION

  Relationship

  OPERATOR_COMMISSIONED_MULTIVARIABLE_EVALUATION

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  Allegheny County Department of Human Services / commissioned Stanford evaluation

  Publication / access

  2023-09-01 / 2026-07-28

  Supports: Reported changes in screening and later child-welfare outcomes after AFST implementation · Risk-stratified and race-stratified analyses · The report’s own methodological and causal limitations

  Does not support: 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

• Algorithms, Humans and Racial Disparities in Child Protection Systems

  Source ID

  AUTH-26-AFST-HUMAN-RACIAL-DISPARITY

  Class

  PEER_REVIEWED_OR_ACADEMIC_EMPIRICAL_STUDY

  Relationship

  ACADEMIC_EMPIRICAL_ANALYSIS

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  Peer-reviewed / academic research associated with AFST

  Publication / access

  2023-06-19 / 2026-07-28

  Supports: Human reviewers did not mechanically reproduce algorithm-only recommendations · Human intervention altered measured Black–white screen-in disparity in the studied setting · Automation effects must be evaluated as a human-system process

  Does not support: That all human review reduces disparities · Complete absence of disparate impact · Generalization beyond the AFST setting and period

• An inspection of visit visa operations, December 2022 to January 2023

  Source ID

  AUTH-27-CARS-ICIBI-INSPECTION

  Class

  INDEPENDENT_GOVERNMENT_INSPECTION

  Relationship

  INDEPENDENT_INSPECTION_OF_OPERATOR_WORKFLOW

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  Independent Chief Inspector of Borders and Immigration

  Publication / access

  2023-04-20 / 2026-07-28

  Supports: CARS appeared generally accurate in routing application complexity in the inspected period · Complex applications took longer and some practices emerged outside policy · The inspection found a lack of first-line assurance and recommended continuing equality review

  Does not support: 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

• Home Office response to inspection of visit visa operations

  Source ID

  AUTH-28-CARS-HOME-OFFICE-RESPONSE

  Class

  OFFICIAL_OPERATOR_RESPONSE

  Relationship

  PRIMARY_OPERATOR_RESPONSE_TO_INDEPENDENT_INSPECTION

  Independence

  INDEPENDENT_FOR_LIMITED_PROPOSITION

  Issuing body

  UK Home Office

  Publication / access

  2023-04-20 / 2026-07-28

  Supports: Operator acceptance of all five inspection recommendations · Permanent implementation status announced in April 2023 · Operator description of equality-profile review

  Does not support: Independent proof that every recommendation produced the intended outcome · Applicant-level fairness or remedy · Current 2026 performance without a new inspection

• FluSight 2024–2025 Season Evaluation

  Source ID

  AUTH-29-FLUSIGHT-2024-25-EVALUATION

  Class

  OFFICIAL_OPERATOR_EVALUATION

  Relationship

  PRIMARY_OPERATOR_PERFORMANCE_EVALUATION

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  U.S. Centers for Disease Control and Prevention

  Publication / access

  2025-09-16 / 2026-07-28

  Supports: The ensemble intervals failed to anticipate the late-December 2024 and January 2025 hospitalization increase · Forecast publication was delayed until March 5, 2025 after prediction intervals reliably captured observed outcomes · Model comparison and uncertainty treatment for the season

  Does not support: A guarantee of future forecast accuracy · A direct causal estimate of planning or health outcomes produced by the forecasts · Individual-level medical decisions

• FluSight 2023–2024 Season Evaluation

  Source ID

  AUTH-30-FLUSIGHT-2023-24-EVALUATION

  Class

  OFFICIAL_OPERATOR_EVALUATION

  Relationship

  PRIMARY_OPERATOR_PERFORMANCE_EVALUATION

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  U.S. Centers for Disease Control and Prevention

  Publication / access

  2025-04-01 / 2026-07-28

  Supports: The ensemble outperformed most submitted models in the season · Rapid changes remained difficult and intervals missed an observed increase and decline · The role of ensemble methods and calibrated uncertainty

  Does not support: Direct evidence of downstream resource-allocation benefit · Performance outside evaluated seasons · Individual clinical validity

• Project Jetson technical specifications

  Source ID

  AUTH-31-UNHCR-JETSON-TECHNICAL

  Class

  OFFICIAL_TECHNICAL_DESCRIPTION

  Relationship

  PRIMARY_OPERATOR_TECHNICAL_DESCRIPTION

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  United Nations High Commissioner for Refugees

  Publication / access

  2018-01-01 / 2026-07-28

  Supports: Historical data covered seven years and 18 Somalia regions · The stated modeling objective and data pipeline · Operator-reported technical limitations

  Does not support: Independent out-of-sample operational impact · Affected-community benefit or harm · Current production deployment across UNHCR

• UNHCR AI Approach

  Source ID

  AUTH-32-UNHCR-AI-APPROACH-2025

  Class

  OFFICIAL_POLICY_AND_PROGRAM_DESCRIPTION

  Relationship

  PRIMARY_OPERATOR_POLICY_DESCRIPTION

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  United Nations High Commissioner for Refugees

  Publication / access

  2025-09-01 / 2026-07-28

  Supports: UNHCR’s stated current institutional approach to predictive analytics and AI · Use cases described for planning, supply chain, nowcasting, and scenario exploration · Governance principles asserted by the operator

  Does not support: Independent model accuracy or benefit · Individual or community consent and remedy outcomes · A single unified production model

• Allegheny County FY 2026–27 Needs-Based Plan and Budget describing continued AFST use

  Source ID

  AUTH-54-AFST-FY2026-27-BUDGET

  Class

  OFFICIAL_CURRENT_PROGRAM_PLAN

  Relationship

  PRIMARY_OPERATOR_CURRENTNESS_RECORD

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  Allegheny County Department of Human Services

  Publication / access

  FY 2026–27 plan; accessed 2026-07-28 / 2026-07-28

  Supports: Continued operator-described use in the FY 2026–27 planning record · The tool’s screening-assistance role · Currentness of the operator’s stated program architecture

  Does not support: Independent current calibration or subgroup validity · Complete notice, appeal, correction, or downstream repair outcomes · Causal improvement in child or family outcomes

• FluSight final forecast release for the 2025–2026 influenza season

  Source ID

  AUTH-55-FLUSIGHT-2025-26-FINAL-RELEASE

  Class

  OFFICIAL_CURRENT_SEASON_STATUS

  Relationship

  PRIMARY_OPERATOR_SEASON_CURRENTNESS_RECORD

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  U.S. Centers for Disease Control and Prevention

  Publication / access

  2026-06-11 / 2026-07-28

  Supports: The program remained operational through the 2025–2026 season · Seasonal rather than continuous publication status · A planned resumption for the 2026–2027 season

  Does not support: Uniform accuracy at every horizon or during rapid trend changes · A person-level risk decision system · Causal effect on any particular public-health intervention

• Home Office response confirming continued use and published guidance for CARS

  Source ID

  AUTH-56-CARS-2026-HOME-OFFICE-RESPONSE

  Class

  OFFICIAL_CURRENT_PROGRAM_RESPONSE

  Relationship

  PRIMARY_OPERATOR_CURRENTNESS_AND_SCOPE_RECORD

  Independence

  NOT_INDEPENDENT_OPERATOR_OR_ISSUER_RECORD

  Issuing body

  UK Home Office

  Publication / access

  2026-06-25 / 2026-07-28

  Supports: Continued use of CARS in 2026 · The operator’s stated routing rather than final-decision role · Current public-guidance status

  Does not support: Independent subgroup error rates or fairness · Absence of automation bias in downstream scrutiny · Complete applicant notice, appeal, or remedy outcomes