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

AI GOVERNANCE · DELEGATED AUTHORITY · DECISION TRACE · REDRESS

AI Governance Accountability Registry

Fifteen source-bounded control patterns show who remains accountable, what an AI system may do, what cannot be delegated, what evidence a human needs, and how correction and appeal must propagate.

Browse 15 patternsJSON v2JSON v1Source states
REAL-WORLD INTERPRETIVE

A registry is not authority.

AI assistance is not legal office-holding. Technical capability is not delegated authority. Delegated workflow is not delegated fiduciary duty. A model recommendation is not a lawful public or corporate act. A municipal dashboard, digital twin, or urban operating system is not democratic legitimacy. A public-company AI claim, ticker, revenue figure, valuation, or market forecast is not investment suitability. A pilot, press release, procurement notice, demonstration, symbolic appointment, or vendor claim is not verified autonomous control. Formal human presence is not meaningful review without knowledge, time, independence, authority, and an effective intervention path.

REAL-WORLD INTERPRETIVE

Prepared controls are not completed review.

Prepared is not executed. Submitted is not accepted. A fixture is not a real reviewer, adjudicator, correction, acknowledgment, migration, participant, or endorsement. Schema-valid, checksum-valid, signature-valid, source-preserved, transport-accepted, or market-listed is not truth, legality, fiduciary compliance, downstream repair, democratic legitimacy, investment suitability, publication approval, production authorization, human accessibility certification, specialist review, civil-rights review, scientific review, clinical review, or operational effectiveness.

JSON

DETERMINISTIC PUBLIC REPRESENTATIONS

Versioned exports with fail-closed schema handling

REAL-WORLD INTERPRETIVE

Schema v2

psychologicalwar.ai-governance-accountability-registry.v2

Records: 15

ec3ed7a47555ffbc083ac412a850038eeebe29723773b89eab2a8bf953ad1515

REAL-WORLD INTERPRETIVE

Compatibility v1

psychologicalwar.ai-governance-accountability-registry.v1

Records: 15

6585f8ba076d212e32e8b5be101dbfed3c8a9d0eb020f4ad65eb8c7e8641fc87

REAL-WORLD INTERPRETIVE

Consumer state

Proven external migrations: 0

Unsupported schemas return HTTP 400 and are not silently coerced.

15

ACCOUNTABILITY PATTERNS

Delegated technical role versus non-delegable authority

15 of 15 patterns shown.

REAL-WORLD INTERPRETIVE

National / public administration

National policy decision support

AIGOV-80-01-national-policy-decision-support

Accountable owner: Named minister, agency head, or lawful decision body

Delegated AI role: Retrieve, summarize, compare scenarios, forecast bounded indicators, and recommend options

Cannot be delegated

Exercise sovereign or statutory authority; adopt policy; order coercive action; or waive legal duties

Authority, review, redress, and currentness
Authority source
Constitution, statute, regulation, delegated administrative authority, or recorded policy mandate
Human evidence view
The accountable human must receive independently intelligible source, uncertainty, alternative, conflict, and consequence information rather than only a ranked recommendation.
Refusal authority
The authorized human must be able to reject, hold, escalate, or request additional evidence without an undocumented penalty or automatic adverse default.
Intervention
A tested pause, rollback, shutdown, credential revocation, or safe-state path must match the consequence and timing of the delegated function.
Notice
Affected people should receive legally and operationally appropriate notice, the responsible institution, the material basis, and the available challenge path.
Appeal
Appeal must reach an authorized human or institution capable of changing the outcome and repairing downstream records.
Correction
Correction closure requires target-specific repair evidence; later contradictory evidence can reopen the record.
Sunset/successor
Delegation expires or is revalidated on material system, vendor, data, authority, purpose, or operating-context change; successor systems inherit unresolved duties.
Evidence state
SOURCE_BOUNDED_ANALYTICAL_CONTROL_MODEL
Currentness
SOURCE_DATE_BOUNDED_REVIEW_REQUIRED
Source
AI Proxy Governance In Nations.md
Digest
c74629f26499d60273479a315edf7b8db0a38a7e0583b5342cdf66c16362d16a
REAL-WORLD INTERPRETIVE

National / regional / municipal

Emergency-management coordination

AIGOV-80-02-emergency-management-coordination

Accountable owner: Authorized emergency-management official or incident command

Delegated AI role: Fuse reports, estimate resource demand, route alerts, and propose response options

Cannot be delegated

Declare an emergency, compel conduct, restrict rights, or authorize extraordinary powers

Authority, review, redress, and currentness
Authority source
Emergency law, incident-command plan, and agency delegation
Human evidence view
The accountable human must receive independently intelligible source, uncertainty, alternative, conflict, and consequence information rather than only a ranked recommendation.
Refusal authority
The authorized human must be able to reject, hold, escalate, or request additional evidence without an undocumented penalty or automatic adverse default.
Intervention
A tested pause, rollback, shutdown, credential revocation, or safe-state path must match the consequence and timing of the delegated function.
Notice
Affected people should receive legally and operationally appropriate notice, the responsible institution, the material basis, and the available challenge path.
Appeal
Appeal must reach an authorized human or institution capable of changing the outcome and repairing downstream records.
Correction
Correction closure requires target-specific repair evidence; later contradictory evidence can reopen the record.
Sunset/successor
Delegation expires or is revalidated on material system, vendor, data, authority, purpose, or operating-context change; successor systems inherit unresolved duties.
Evidence state
SOURCE_BOUNDED_ANALYTICAL_CONTROL_MODEL
Currentness
SOURCE_DATE_BOUNDED_REVIEW_REQUIRED
Source
AI Proxy Governance In Nations.md
Digest
cf518d5dc77523be48dc1ba91fd79e0d6adfff81a078bb6c8322717f1358f003
REAL-WORLD INTERPRETIVE

Municipal administration

Municipal benefits or service triage

AIGOV-80-03-municipal-benefits-triage

Accountable owner: Named department director and authorized case decision-maker

Delegated AI role: Classify requests, prioritize queues, identify missing records, and draft explanations

Cannot be delegated

Determine legal eligibility, terminate benefits, impose debt, or deny due process without lawful human decision

Authority, review, redress, and currentness
Authority source
Municipal code, benefits law, service rule, or approved administrative procedure
Human evidence view
The accountable human must receive independently intelligible source, uncertainty, alternative, conflict, and consequence information rather than only a ranked recommendation.
Refusal authority
The authorized human must be able to reject, hold, escalate, or request additional evidence without an undocumented penalty or automatic adverse default.
Intervention
A tested pause, rollback, shutdown, credential revocation, or safe-state path must match the consequence and timing of the delegated function.
Notice
Affected people should receive legally and operationally appropriate notice, the responsible institution, the material basis, and the available challenge path.
Appeal
Appeal must reach an authorized human or institution capable of changing the outcome and repairing downstream records.
Correction
Correction closure requires target-specific repair evidence; later contradictory evidence can reopen the record.
Sunset/successor
Delegation expires or is revalidated on material system, vendor, data, authority, purpose, or operating-context change; successor systems inherit unresolved duties.
Evidence state
SOURCE_BOUNDED_ANALYTICAL_CONTROL_MODEL
Currentness
SOURCE_DATE_BOUNDED_REVIEW_REQUIRED
Source
AI in Municipal Governance.md
Digest
054388f5ace013dda1fccc911b66ddf03131a265ebc6f06c26948dc37d67d3d5
REAL-WORLD INTERPRETIVE

Municipal infrastructure

Municipal infrastructure digital twin

AIGOV-80-04-municipal-infrastructure-digital-twin

Accountable owner: Named public-works authority and accountable vendor owner

Delegated AI role: Model infrastructure state, predict maintenance needs, and recommend scheduling

Cannot be delegated

Authorize unsafe operation, conceal uncertainty, override safety law, or replace accountable engineering judgment

Authority, review, redress, and currentness
Authority source
Procurement contract, safety code, operational policy, and public authority
Human evidence view
The accountable human must receive independently intelligible source, uncertainty, alternative, conflict, and consequence information rather than only a ranked recommendation.
Refusal authority
The authorized human must be able to reject, hold, escalate, or request additional evidence without an undocumented penalty or automatic adverse default.
Intervention
A tested pause, rollback, shutdown, credential revocation, or safe-state path must match the consequence and timing of the delegated function.
Notice
Affected people should receive legally and operationally appropriate notice, the responsible institution, the material basis, and the available challenge path.
Appeal
Appeal must reach an authorized human or institution capable of changing the outcome and repairing downstream records.
Correction
Correction closure requires target-specific repair evidence; later contradictory evidence can reopen the record.
Sunset/successor
Delegation expires or is revalidated on material system, vendor, data, authority, purpose, or operating-context change; successor systems inherit unresolved duties.
Evidence state
SOURCE_BOUNDED_ANALYTICAL_CONTROL_MODEL
Currentness
SOURCE_DATE_BOUNDED_REVIEW_REQUIRED
Source
AI in Municipal Governance.md
Digest
229d177a137460c5631dbaf41a6b39d13a0ce1149291f9f782577f881b800d54
REAL-WORLD INTERPRETIVE

Municipal public safety

Municipal public-safety analytics

AIGOV-80-05-municipal-public-safety-analytics

Accountable owner: Named public-safety official with legally bounded authority

Delegated AI role: Map incidents, correlate records, prioritize review, and recommend nonbinding resource options

Cannot be delegated

Create guilt, probable cause, warrant authority, or coercive power merely from a score or alert

Authority, review, redress, and currentness
Authority source
Public-safety law, departmental policy, civil-rights limits, and documented purpose
Human evidence view
The accountable human must receive independently intelligible source, uncertainty, alternative, conflict, and consequence information rather than only a ranked recommendation.
Refusal authority
The authorized human must be able to reject, hold, escalate, or request additional evidence without an undocumented penalty or automatic adverse default.
Intervention
A tested pause, rollback, shutdown, credential revocation, or safe-state path must match the consequence and timing of the delegated function.
Notice
Affected people should receive legally and operationally appropriate notice, the responsible institution, the material basis, and the available challenge path.
Appeal
Appeal must reach an authorized human or institution capable of changing the outcome and repairing downstream records.
Correction
Correction closure requires target-specific repair evidence; later contradictory evidence can reopen the record.
Sunset/successor
Delegation expires or is revalidated on material system, vendor, data, authority, purpose, or operating-context change; successor systems inherit unresolved duties.
Evidence state
SOURCE_BOUNDED_ANALYTICAL_CONTROL_MODEL
Currentness
SOURCE_DATE_BOUNDED_REVIEW_REQUIRED
Source
AI in Municipal Governance.md
Digest
2736813614d79f6bea8f8aa90c93aed8b7ece11c4e8a2c737cb50bb0bc2a1750
REAL-WORLD INTERPRETIVE

Public procurement

Public procurement vendor scoring

AIGOV-80-06-public-procurement-vendor-scoring

Accountable owner: Authorized procurement official and review board

Delegated AI role: Check documents, compare criteria, flag inconsistencies, and propose rankings

Cannot be delegated

Award a contract, waive mandatory criteria, hide conflicts, or deny protest rights

Authority, review, redress, and currentness
Authority source
Procurement law, bid criteria, conflict rules, and appeal procedure
Human evidence view
The accountable human must receive independently intelligible source, uncertainty, alternative, conflict, and consequence information rather than only a ranked recommendation.
Refusal authority
The authorized human must be able to reject, hold, escalate, or request additional evidence without an undocumented penalty or automatic adverse default.
Intervention
A tested pause, rollback, shutdown, credential revocation, or safe-state path must match the consequence and timing of the delegated function.
Notice
Affected people should receive legally and operationally appropriate notice, the responsible institution, the material basis, and the available challenge path.
Appeal
Appeal must reach an authorized human or institution capable of changing the outcome and repairing downstream records.
Correction
Correction closure requires target-specific repair evidence; later contradictory evidence can reopen the record.
Sunset/successor
Delegation expires or is revalidated on material system, vendor, data, authority, purpose, or operating-context change; successor systems inherit unresolved duties.
Evidence state
SOURCE_BOUNDED_ANALYTICAL_CONTROL_MODEL
Currentness
SOURCE_DATE_BOUNDED_REVIEW_REQUIRED
Source
AI Proxy Governance In Nations.md
Digest
27ea72280ba588e21d60806f6c8258dd10c02edf6f9da620c657db108b9c354f
REAL-WORLD INTERPRETIVE

Corporate governance

Corporate board decision support

AIGOV-80-07-corporate-board-decision-support

Accountable owner: Human directors and board committees

Delegated AI role: Analyze materials, model scenarios, identify risks, and draft recommendations

Cannot be delegated

Serve as a legal director, discharge fiduciary duty, vote independently, or absorb liability

Authority, review, redress, and currentness
Authority source
Corporate law, charter, bylaws, fiduciary duties, and board resolutions
Human evidence view
The accountable human must receive independently intelligible source, uncertainty, alternative, conflict, and consequence information rather than only a ranked recommendation.
Refusal authority
The authorized human must be able to reject, hold, escalate, or request additional evidence without an undocumented penalty or automatic adverse default.
Intervention
A tested pause, rollback, shutdown, credential revocation, or safe-state path must match the consequence and timing of the delegated function.
Notice
Affected people should receive legally and operationally appropriate notice, the responsible institution, the material basis, and the available challenge path.
Appeal
Appeal must reach an authorized human or institution capable of changing the outcome and repairing downstream records.
Correction
Correction closure requires target-specific repair evidence; later contradictory evidence can reopen the record.
Sunset/successor
Delegation expires or is revalidated on material system, vendor, data, authority, purpose, or operating-context change; successor systems inherit unresolved duties.
Evidence state
SOURCE_BOUNDED_ANALYTICAL_CONTROL_MODEL
Currentness
SOURCE_DATE_BOUNDED_REVIEW_REQUIRED
Source
AI Proxy Corporate Leadership.md
Digest
842d1bac8bec27fd38e4d057bc03476da0e434e56bce80486b6fcc3a372c84f7
REAL-WORLD INTERPRETIVE

Corporate operations

Algorithmic executive workflow

AIGOV-80-08-algorithmic-executive-workflow

Accountable owner: Human officers, directors, and designated control owners

Delegated AI role: Plan tasks, route work, propose pricing, allocate resources, and monitor operations

Cannot be delegated

Become a lawful officer, bind the company outside delegated authority, or replace accountable human judgment

Authority, review, redress, and currentness
Authority source
Corporate authority matrix, delegated operating policy, and fiduciary obligations
Human evidence view
The accountable human must receive independently intelligible source, uncertainty, alternative, conflict, and consequence information rather than only a ranked recommendation.
Refusal authority
The authorized human must be able to reject, hold, escalate, or request additional evidence without an undocumented penalty or automatic adverse default.
Intervention
A tested pause, rollback, shutdown, credential revocation, or safe-state path must match the consequence and timing of the delegated function.
Notice
Affected people should receive legally and operationally appropriate notice, the responsible institution, the material basis, and the available challenge path.
Appeal
Appeal must reach an authorized human or institution capable of changing the outcome and repairing downstream records.
Correction
Correction closure requires target-specific repair evidence; later contradictory evidence can reopen the record.
Sunset/successor
Delegation expires or is revalidated on material system, vendor, data, authority, purpose, or operating-context change; successor systems inherit unresolved duties.
Evidence state
SOURCE_BOUNDED_ANALYTICAL_CONTROL_MODEL
Currentness
SOURCE_DATE_BOUNDED_REVIEW_REQUIRED
Source
AI Proxy Corporate Leadership.md
Digest
f275f4ec3c01e9d8c9c25b6d2799fe2788ec4d33356bd9b1b774a139143cfeee
REAL-WORLD INTERPRETIVE

Autonomous enterprise

Autonomous-enterprise operations

AIGOV-80-09-autonomous-enterprise-operations

Accountable owner: Human owner, manager, director, or legally recognized entity controller

Delegated AI role: Orchestrate agents, use bounded tools, manage routine commerce, and propose operational changes

Cannot be delegated

Own property as an AI, evade beneficial-owner duties, self-expand authority, or make unreviewable high-impact decisions

Authority, review, redress, and currentness
Authority source
Entity law, account authority, contracts, banking controls, and operating policy
Human evidence view
The accountable human must receive independently intelligible source, uncertainty, alternative, conflict, and consequence information rather than only a ranked recommendation.
Refusal authority
The authorized human must be able to reject, hold, escalate, or request additional evidence without an undocumented penalty or automatic adverse default.
Intervention
A tested pause, rollback, shutdown, credential revocation, or safe-state path must match the consequence and timing of the delegated function.
Notice
Affected people should receive legally and operationally appropriate notice, the responsible institution, the material basis, and the available challenge path.
Appeal
Appeal must reach an authorized human or institution capable of changing the outcome and repairing downstream records.
Correction
Correction closure requires target-specific repair evidence; later contradictory evidence can reopen the record.
Sunset/successor
Delegation expires or is revalidated on material system, vendor, data, authority, purpose, or operating-context change; successor systems inherit unresolved duties.
Evidence state
SOURCE_BOUNDED_ANALYTICAL_CONTROL_MODEL
Currentness
SOURCE_DATE_BOUNDED_REVIEW_REQUIRED
Source
AI Owned and Run Businesses.md
Digest
05683eabd56bea54814860f8541f24adaa2f1bb9e23050966404c340786f6116
REAL-WORLD INTERPRETIVE

Finance / insurance

Financial underwriting and risk pricing

AIGOV-80-10-financial-underwriting-and-pricing

Accountable owner: Licensed institution and named accountable officers

Delegated AI role: Extract factors, estimate risk, prioritize review, and recommend pricing within bounded rules

Cannot be delegated

Create lawful discrimination, conceal adverse-action reasons, or transfer compliance obligations to a model

Authority, review, redress, and currentness
Authority source
Financial law, underwriting policy, model-risk governance, and fair-treatment duties
Human evidence view
The accountable human must receive independently intelligible source, uncertainty, alternative, conflict, and consequence information rather than only a ranked recommendation.
Refusal authority
The authorized human must be able to reject, hold, escalate, or request additional evidence without an undocumented penalty or automatic adverse default.
Intervention
A tested pause, rollback, shutdown, credential revocation, or safe-state path must match the consequence and timing of the delegated function.
Notice
Affected people should receive legally and operationally appropriate notice, the responsible institution, the material basis, and the available challenge path.
Appeal
Appeal must reach an authorized human or institution capable of changing the outcome and repairing downstream records.
Correction
Correction closure requires target-specific repair evidence; later contradictory evidence can reopen the record.
Sunset/successor
Delegation expires or is revalidated on material system, vendor, data, authority, purpose, or operating-context change; successor systems inherit unresolved duties.
Evidence state
SOURCE_BOUNDED_ANALYTICAL_CONTROL_MODEL
Currentness
SOURCE_DATE_BOUNDED_REVIEW_REQUIRED
Source
Publicly Traded AI Companies.md
Digest
a79240c83251ee5b4883a37fea14678d16360f1fc74ca5b839c6624b67344377
REAL-WORLD INTERPRETIVE

Public markets

Public-company AI disclosure and claims

AIGOV-80-11-public-company-ai-disclosure

Accountable owner: Issuer officers, disclosure committee, auditors, and board

Delegated AI role: Inventory AI systems, draft disclosures, compare metrics, and flag inconsistencies

Cannot be delegated

Certify material accuracy, establish valuation, or provide investment suitability

Authority, review, redress, and currentness
Authority source
Securities law, accounting rules, listing obligations, and disclosure controls
Human evidence view
The accountable human must receive independently intelligible source, uncertainty, alternative, conflict, and consequence information rather than only a ranked recommendation.
Refusal authority
The authorized human must be able to reject, hold, escalate, or request additional evidence without an undocumented penalty or automatic adverse default.
Intervention
A tested pause, rollback, shutdown, credential revocation, or safe-state path must match the consequence and timing of the delegated function.
Notice
Affected people should receive legally and operationally appropriate notice, the responsible institution, the material basis, and the available challenge path.
Appeal
Appeal must reach an authorized human or institution capable of changing the outcome and repairing downstream records.
Correction
Correction closure requires target-specific repair evidence; later contradictory evidence can reopen the record.
Sunset/successor
Delegation expires or is revalidated on material system, vendor, data, authority, purpose, or operating-context change; successor systems inherit unresolved duties.
Evidence state
SOURCE_BOUNDED_ANALYTICAL_CONTROL_MODEL
Currentness
SOURCE_DATE_BOUNDED_REVIEW_REQUIRED
Source
Publicly Traded AI Companies.md
Digest
412c2c2cd0d0d6f9e28d04ae7dd2e9527796ac7233648e67945c9142f6e387f5
REAL-WORLD INTERPRETIVE

Employment

Employment scheduling and performance management

AIGOV-80-12-employment-algorithmic-management

Accountable owner: Employer, named managers, and authorized human-resources decision-makers

Delegated AI role: Schedule work, summarize performance signals, and recommend review

Cannot be delegated

Terminate, discipline, deny accommodation, or evade labor rights solely through opaque automation

Authority, review, redress, and currentness
Authority source
Employment law, collective agreements, workplace policy, and accommodations duties
Human evidence view
The accountable human must receive independently intelligible source, uncertainty, alternative, conflict, and consequence information rather than only a ranked recommendation.
Refusal authority
The authorized human must be able to reject, hold, escalate, or request additional evidence without an undocumented penalty or automatic adverse default.
Intervention
A tested pause, rollback, shutdown, credential revocation, or safe-state path must match the consequence and timing of the delegated function.
Notice
Affected people should receive legally and operationally appropriate notice, the responsible institution, the material basis, and the available challenge path.
Appeal
Appeal must reach an authorized human or institution capable of changing the outcome and repairing downstream records.
Correction
Correction closure requires target-specific repair evidence; later contradictory evidence can reopen the record.
Sunset/successor
Delegation expires or is revalidated on material system, vendor, data, authority, purpose, or operating-context change; successor systems inherit unresolved duties.
Evidence state
SOURCE_BOUNDED_ANALYTICAL_CONTROL_MODEL
Currentness
SOURCE_DATE_BOUNDED_REVIEW_REQUIRED
Source
AI Proxy Corporate Leadership.md
Digest
14f5ccdc060c8d216752006f09121470bf029f31029f9eb87517f13b6f9b4f8e
REAL-WORLD INTERPRETIVE

Consumer operations

Customer service and complaint resolution

AIGOV-80-13-customer-service-and-complaint-resolution

Accountable owner: Named service owner and authorized complaint adjudicator

Delegated AI role: Answer routine questions, retrieve records, classify complaints, and propose resolutions

Cannot be delegated

Waive legal rights, make binding settlements, or close serious complaints without accountable review

Authority, review, redress, and currentness
Authority source
Consumer law, service terms, complaint policy, and escalation procedure
Human evidence view
The accountable human must receive independently intelligible source, uncertainty, alternative, conflict, and consequence information rather than only a ranked recommendation.
Refusal authority
The authorized human must be able to reject, hold, escalate, or request additional evidence without an undocumented penalty or automatic adverse default.
Intervention
A tested pause, rollback, shutdown, credential revocation, or safe-state path must match the consequence and timing of the delegated function.
Notice
Affected people should receive legally and operationally appropriate notice, the responsible institution, the material basis, and the available challenge path.
Appeal
Appeal must reach an authorized human or institution capable of changing the outcome and repairing downstream records.
Correction
Correction closure requires target-specific repair evidence; later contradictory evidence can reopen the record.
Sunset/successor
Delegation expires or is revalidated on material system, vendor, data, authority, purpose, or operating-context change; successor systems inherit unresolved duties.
Evidence state
SOURCE_BOUNDED_ANALYTICAL_CONTROL_MODEL
Currentness
SOURCE_DATE_BOUNDED_REVIEW_REQUIRED
Source
AI Owned and Run Businesses.md
Digest
41f24b71f343946985cc8080be373cf9495e7b3473fa0d81a39e3099433d8932
REAL-WORLD INTERPRETIVE

Cybersecurity

Cybersecurity autonomous response

AIGOV-80-14-cybersecurity-autonomous-response

Accountable owner: Named security owner and incident commander

Delegated AI role: Detect anomalies, isolate bounded assets, rotate credentials, and recommend containment

Cannot be delegated

Expand privileges, destroy evidence, affect third parties, or execute irreversible action outside preauthorized bounds

Authority, review, redress, and currentness
Authority source
Security policy, incident plan, access-control rules, and contractual authority
Human evidence view
The accountable human must receive independently intelligible source, uncertainty, alternative, conflict, and consequence information rather than only a ranked recommendation.
Refusal authority
The authorized human must be able to reject, hold, escalate, or request additional evidence without an undocumented penalty or automatic adverse default.
Intervention
A tested pause, rollback, shutdown, credential revocation, or safe-state path must match the consequence and timing of the delegated function.
Notice
Affected people should receive legally and operationally appropriate notice, the responsible institution, the material basis, and the available challenge path.
Appeal
Appeal must reach an authorized human or institution capable of changing the outcome and repairing downstream records.
Correction
Correction closure requires target-specific repair evidence; later contradictory evidence can reopen the record.
Sunset/successor
Delegation expires or is revalidated on material system, vendor, data, authority, purpose, or operating-context change; successor systems inherit unresolved duties.
Evidence state
SOURCE_BOUNDED_ANALYTICAL_CONTROL_MODEL
Currentness
SOURCE_DATE_BOUNDED_REVIEW_REQUIRED
Source
AI Owned and Run Businesses.md
Digest
02e4005a4a454cf305a65a8d84be1d40322557349288ae0b47cb19bd0d0c8a8a
REAL-WORLD INTERPRETIVE

Political communication / public information

Synthetic political or public communication actor

AIGOV-80-15-synthetic-political-or-public-communication

Accountable owner: Named campaign, party, public office, or institutional publisher

Delegated AI role: Draft, translate, personalize, schedule, and analyze communications

Cannot be delegated

Become a democratic representative, fabricate authority, conceal synthetic identity where disclosure is required, or make official decisions

Authority, review, redress, and currentness
Authority source
Election law, public-information authority, platform rules, and attribution policy
Human evidence view
The accountable human must receive independently intelligible source, uncertainty, alternative, conflict, and consequence information rather than only a ranked recommendation.
Refusal authority
The authorized human must be able to reject, hold, escalate, or request additional evidence without an undocumented penalty or automatic adverse default.
Intervention
A tested pause, rollback, shutdown, credential revocation, or safe-state path must match the consequence and timing of the delegated function.
Notice
Affected people should receive legally and operationally appropriate notice, the responsible institution, the material basis, and the available challenge path.
Appeal
Appeal must reach an authorized human or institution capable of changing the outcome and repairing downstream records.
Correction
Correction closure requires target-specific repair evidence; later contradictory evidence can reopen the record.
Sunset/successor
Delegation expires or is revalidated on material system, vendor, data, authority, purpose, or operating-context change; successor systems inherit unresolved duties.
Evidence state
SOURCE_BOUNDED_ANALYTICAL_CONTROL_MODEL
Currentness
SOURCE_DATE_BOUNDED_REVIEW_REQUIRED
Source
AI Proxy Governance In Nations.md
Digest
b0baffcd80755000b202d489348c7737bf74802d7c231538e4d7ddda824e4636
16

MEANINGFUL AUTHORITY CONTROLS

Human presence is insufficient without enforceable conditions

ControlRequirementState
named-accountable-ownerA named human or institution owns the decision and remedy obligations.REQUIRED_NOT_EXTERNALLY_CERTIFIED
documented-authority-basisThe legal, corporate, contractual, or policy authority is recorded and current.REQUIRED_NOT_EXTERNALLY_CERTIFIED
bounded-purpose-and-scopePurpose, population, geography, time, data, tools, and prohibited uses are explicit.REQUIRED_NOT_EXTERNALLY_CERTIFIED
source-and-data-provenanceMaterial inputs retain origin, time, transformation, access, and integrity provenance.REQUIRED_NOT_EXTERNALLY_CERTIFIED
model-rule-and-limit-disclosureThe system role, rule layer, model limitations, uncertainty, and abstention conditions are visible to reviewers.REQUIRED_NOT_EXTERNALLY_CERTIFIED
independent-evidence-viewThe reviewer can inspect material evidence and alternatives without depending solely on the model summary.REQUIRED_NOT_EXTERNALLY_CERTIFIED
adequate-time-and-workloadTime, queue size, staffing, and interface design permit actual scrutiny.REQUIRED_NOT_EXTERNALLY_CERTIFIED
refusal-and-escalation-authorityThe reviewer can reject, hold, escalate, or seek evidence without automatic adverse action.REQUIRED_NOT_EXTERNALLY_CERTIFIED
effective-technical-interventionPause, rollback, shutdown, credential revocation, and safe-state controls are tested under realistic latency.REQUIRED_NOT_EXTERNALLY_CERTIFIED
segregation-of-dutiesModel owner, operator, adjudicator, auditor, and correction authority are separated where risk requires.REQUIRED_NOT_EXTERNALLY_CERTIFIED
attributable-identity-and-key-custodyHuman and service actions are attributable; keys and privileged identities have accountable custody.REQUIRED_NOT_EXTERNALLY_CERTIFIED
dissent-incident-and-near-miss-preservationDissent, overrides, incidents, near misses, and rejected recommendations remain auditable.REQUIRED_NOT_EXTERNALLY_CERTIFIED
notice-and-explanationAffected people receive the responsible institution, material basis, limits, and available next steps when law and safety permit.REQUIRED_NOT_EXTERNALLY_CERTIFIED
appeal-and-redressAn appeal reaches a human or institution with authority and resources to change the result.REQUIRED_NOT_EXTERNALLY_CERTIFIED
correction-propagationAccepted correction repairs every applicable canonical, public, memory, cache, derived, consumer, and successor target.REQUIRED_NOT_EXTERNALLY_CERTIFIED
sunset-successor-and-revalidationAuthority and controls expire or are revalidated after material changes; successor systems inherit unresolved obligations.REQUIRED_NOT_EXTERNALLY_CERTIFIED
18

END-TO-END DECISION TRACE

Trace authority from source through sunset

  1. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-01

    source-acquisition

    Acquire an authorized source and record provenance.

  2. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-02

    legal-or-organizational-basis

    Confirm lawful or organizational basis before processing.

  3. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-03

    identity-and-entity-resolution

    Resolve the subject or entity while preserving ambiguity and mismatch risk.

  4. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-04

    data-quality-and-representativeness

    Assess completeness, accuracy, timeliness, subgroup effects, and missingness.

  5. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-05

    feature-or-rule-construction

    Document transformations, proxies, thresholds, and policy assumptions.

  6. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-06

    model-or-rule-selection

    Record the model, rule, version, owner, intended use, and excluded use.

  7. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-07

    inference-or-recommendation

    Produce a bounded output without silently converting it into authority.

  8. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-08

    confidence-uncertainty-and-alternatives

    Expose calibration, uncertainty, alternative hypotheses, stale evidence, and abstention.

  9. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-09

    human-evidence-view

    Present independently intelligible evidence, not only a score or summary.

  10. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-10

    conflict-and-compensation-disclosure

    Disclose reviewer, vendor, institutional, and compensation conflicts.

  11. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-11

    authority-and-scope-check

    Verify the decision-maker, authority, purpose, scope, and expiration.

  12. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-12

    approval-refusal-or-escalation

    Record the human decision, dissent, requested evidence, and authority used.

  13. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-13

    execution

    Execute only through bounded tools and record the exact external state change.

  14. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-14

    monitoring-and-incident-response

    Monitor outcomes, drift, security, overrides, incidents, and near misses.

  15. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-15

    affected-person-notice

    Provide notice and explanation where lawful and safe.

  16. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-16

    appeal-and-redress

    Receive and adjudicate challenges through an authorized human process.

  17. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-17

    correction-propagation

    Repair every applicable target and prove closure separately.

  18. REAL-WORLD INTERPRETIVE

    AIGOV-TRACE-80-18

    sunset-successor-and-retirement

    Expire, retire, migrate, or revalidate the system without losing unresolved duties.

18

TARGET-SPECIFIC REPAIR

A correction notice is not downstream closure

Target classDefinitionStateClosure boundary
canonical-accountability-recordCanonical registry row and record digestPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
public-accountability-htmlHuman-readable registry pagePREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
schema-v2-exportDefault public JSON representationPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
schema-v1-exportCompatibility JSON representationPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
source-state-exportPublic source-state JSONPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
content-indexPublic content-index inventoryPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
sitemapCanonical sitemapPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
structured-dataJSON-LD and metadataPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
social-metadataOpen Graph and social previewPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
search-indexInternal search indexPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
learning-pathsLearning paths and cross-topic routesPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
durable-reportAuthored durable integration reportPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
protected-uaiProtected domain memoryPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
hot-uaiRoot first-load memoryPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
semantic-pointer-ledgerDurable semantic pointers and checksumsPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
cache-or-content-deliveryCaches and public delivery layersPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
downstream-consumerAuthorized downstream consumer with direct acknowledgmentPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
archive-or-successor-systemArchived snapshot, vendor migration, or successor systemPREPARED_NOT_TRIGGEREDTransport, signature, or cache refresh alone cannot close.
12

ATTRIBUTABLE REVIEW ENVELOPE

Private identity compartment, public-safe receipt, and zero automatic promotion

The contract contains 40 required fields. It keeps identity, consent, attribution, conflicts, compensation, and contact data in bounded privacy classes.

RECEIVED_UNVERIFIEDSCHEMA_INVALIDIDENTITY_UNRESOLVEDAUTHORITY_UNRESOLVEDCONFLICT_DISCLOSURE_INCOMPLETESOURCE_UNAVAILABLEREADY_FOR_TECHNICAL_REVIEWREADY_FOR_SUBSTANTIVE_REVIEWADJUDICATED_NO_CHANGEACCEPTED_WITH_LIMITSACCEPTED_CANONICAL_DELTAWITHDRAWN_OR_REJECTED

Synthetic fixtures: 20. Real submissions: 0.

Page complete AI Governance Accountability Registry Page label: REAL-WORLD INTERPRETIVE