INV-066 — AI Representation Requires Continuous Auditability to the Represented Party

1. Definition

A representation system that cannot be audited by the represented party is not legitimate representation.

AI representation occurs when an AI system acts for, speaks for, models, mirrors, summarizes, simulates, negotiates for, remembers for, decides for, recommends for, ranks for, or presents a person, group, institution, community, role, identity, brand, archive, or public.

AI representation is not merely output generation.

It is a boundary-bearing proxy relationship.

Representation affects:

identity
meaning
agency
memory
consent
authority
reputation
decision rights
context continuity
public perception
contractual action
social trust
legal / economic standing

Therefore:

AI representation requires continuous auditability to the represented party.

If the represented party cannot inspect, correct, scope, revoke, appeal, or exit the representation, the representation is not legitimate.

2. Purpose

This invariant prevents AI from becoming an unaccountable proxy for a person, group, institution, or meaning structure.

AI can represent users in many ways:

• drafting messages
• summarizing views
• maintaining memory
• negotiating
• filling forms
• acting as an agent
• speaking in a user’s voice
• generating avatars
• simulating a persona
• managing calendar or email
• representing a company
• modeling a group preference
• answering on behalf of a user
• producing synthetic likeness
• acting as AI twin or mirror
• mediating public identity
• preserving project canon

The false assumption is:

If the AI is useful or accurate enough, it may represent the party.

The UTS correction is:

Representation remains legitimate only while the represented party can continuously audit and correct it.

Representation without audit becomes capture.

Representation without scope becomes overreach.

Representation without correction becomes identity drift.

Representation without exit becomes coercive proxy.

Representation without memory integrity becomes misrepresentation.

3. Constraint Statement

Canonical Form

AI representation requires continuous auditability to the represented party.

Expanded Form

Any AI system that represents a person, group, institution, role, identity,
archive, public, or meaning structure must remain continuously auditable,
correctable, scoped, revocable, and appealable by the represented party,
with clear boundaries, traceability, rollback, and restoration pathways.

Minimal Expression

No representation without audit.

Representation Form

The represented party must be able to inspect and correct the representation.

AI Governance Form

AI acting for a party requires scope clarity, consent, traceability, correction, rollback, and exit.

Memory Form

AI memory used for representation must remain inspectable, corrigible, and scoped.

Security Form

Representation systems must prevent impersonation, unauthorized agency, and proxy capture.

CMS / Meaning Form

AI may mirror meaning only while meaning remains accountable to the source being represented.

4. Structural Logic

Representation is a coupling between:

represented party
        ↔
AI proxy / model / agent / memory / interface
        ↔
audience, system, or action environment

The AI becomes a mediating layer.

If that layer cannot be audited by the represented party, then the represented party loses control over how their meaning, identity, intent, memory, or authority is expressed.

The incoherent sequence:

AI representation begins
        ↓
representation becomes useful
        ↓
scope expands
        ↓
memory and persona assumptions accumulate
        ↓
represented party loses audit or correction access
        ↓
AI proxy drifts
        ↓
misrepresentation becomes durable
        ↓
identity / agency debt accumulates

The coherent sequence:

AI representation begins
        ↓
scope is explicit
        ↓
consent is valid
        ↓
audit interface remains available
        ↓
represented party can inspect, correct, revoke, and export
        ↓
memory and representation update
        ↓
misrepresentation routes to restoration
        ↓
agency remains preserved

Core insight:

Representation is not legitimate unless the represented party remains sovereign over the proxy.

AI representation must preserve the represented party’s boundary and meaning integrity.

5. State-Vector Impact

Protected State Variables

O   — coherence
Au  — auditability
µᵢ  — meaning / agent integrity
BΣ  — boundary integrity
K   — compatibility between proxy and represented party
R   — restoration capacity
H   — hidden debt

Primary Risk Variables

ι   — inversion when a proxy is mistaken for the represented party
ε   — visible misrepresentation, impersonation, drift, mistaken action, trust failure
Φ   — convenience, productivity, likeness accuracy, automation, engagement, delegation proxy

Healthy Representation Pattern

scope explicit
consent valid
Au continuous
memory corrigible
correction available
rollback available
exit viable
µᵢ preserved
BΣ preserved
O stable or ↑

Violation Pattern

AI proxy active
Au to represented party↓
scope drift↑
memory drift↑
correction unavailable
exit weak
µᵢ↓
BΣ↓
H↑
ι↑
O↓

Proxy-Capture Pattern

AI representation Φ↑
represented-party control↓
proxy authority↑
µᵢ↓
ι↑

The key inversion:

the representation becomes more legible than the represented party.

This is representation capture.

6. U-Layer Localization

Primary Layer

U2 — Configuration / Boundaries

Representation is boundary-governed. Scope, consent, identity, authority, access, and proxy limits are U2 issues.

Memory Layer

U7 — Memory / Recurrence

AI representation often depends on memory. Representation drift becomes recurrent if memory cannot be corrected.

Classification Layer

U4 — Classification / Metrics

AI representation often classifies identity, intent, preference, voice, meaning, role, or context.

Execution Layer

U3 — Execution

AI representation becomes consequential when it sends messages, makes decisions, fills forms, negotiates, publishes, transacts, or acts.

Coherence Field Layer

U6 — Coherence Field

Representation affects trust, reputation, legitimacy, meaning, social recognition, and public perception.

Coordination Layer

U5 — Coordination / Time

Representation must remain valid over time. Old consent, old memory, or old context can become invalid.

Resource Layer

U1 — Power / Budgets

Audit, correction, appeal, rollback, and support require capacity.

Environment Layer

U8 — Environment / Forcing

Market pressure, convenience, automation incentives, social pressure, and platform dependency can push representation beyond valid scope.

Common Failure Pattern

U8 automation incentive
        ↓
AI proxy expands
        ↓
U2 scope blurs
        ↓
U7 memory accumulates assumptions
        ↓
U4 representation categories harden
        ↓
represented party loses audit capacity
        ↓
U6 meaning / trust distorts
        ↓
H↑

Common Misdiagnosis

Representation failure is often misdiagnosed as:

• user misunderstanding
• harmless personalization
• model style issue
• small memory error
• convenience tradeoff
• branding issue
• impersonation edge case
• consent already granted
• automation error
• UX problem
• data quality issue

The deeper issue may be:

The represented party lost auditability over the representation.

7. Violation Signatures

7.1 AI Speaks for a Party Without Review

The AI generates messages, statements, summaries, or positions as if they represent the party, without review or correction.

representation↑
review↓
µᵢ risk↑

7.2 Memory Drift Becomes Representation Drift

Incorrect or outdated memory causes the AI to represent the user, group, or project inaccurately.

memory drift↑
representation drift↑
H↑

7.3 Scope Creep in Delegation

The AI begins with bounded assistance but expands into decisions, commitments, or statements beyond the original scope.

delegation scope↑
consent clarity↓
BΣ↓

7.4 Proxy Acts Without Traceability

The AI acts for a party, but it is unclear what action was taken, why, under what instruction, or with what authority.

proxy action↑
traceability↓
Au↓

7.5 Represented Party Cannot Correct the Record

The represented party cannot edit, revoke, correct, or contextualize the AI’s representation.

misrepresentation detected
correction unavailable
H↑

7.6 Synthetic Likeness Without Ongoing Control

An AI avatar, voice, persona, or twin continues to represent someone without ongoing consent and audit.

likeness Φ↑
control↓
identity debt↑

7.7 Group Representation Without Group Truth Pathways

AI claims to represent a group, community, demographic, institution, or public without affected-node input or correction.

group representation↑
group truth pathways↓
legitimacy debt↑

7.8 AI Representation Used as Authority

The AI-generated representation is treated as more official, concise, searchable, or authoritative than the represented party’s own correction.

proxy legibility↑
source authority↓
ι↑

7.9 Platform Owns the Representation Layer

A platform controls the memory, avatar, ranking, voice, or persona layer through which a person or organization is recognized.

platform representation control↑
exit / portability↓
capture↑

7.10 Project Canon Misrepresented by AI Summary

AI summarizes a canon, archive, or framework incorrectly, then the summary becomes reference material without correction pathway.

AI summary↑
canon accuracy↓
archive H↑

8. Related Failure Modes

Primary related failure modes:

• AI Representation Capture
• Proxy Drift
• Memory-Based Misrepresentation
• Scope Creep in Delegation
• Unauthorized Agency
• Synthetic Likeness Capture
• Persona Drift
• AI Twin Capture
• Group Representation Without Truth
• Proxy Authority Inversion
• Representation Without Auditability
• Correction Path Failure
• Consent Drift
• Memory Non-Corrigibility
• Identity Boundary Failure
• Platform Representation Capture
• Impersonation Risk
• Public Cognition Misrepresentation
• Canon Misrepresentation
• Meaning Integrity Loss
• Hidden Debt Accumulation
• Legitimacy Debt
• Restoration Capacity Lag

9. Related Restoration Arcs

Primary restoration arcs:

• Representation Audit Restoration
• Memory Correction
• Scope Reconstitution
• Consent Renewal
• Proxy Authority Rollback
• Correction Pathway Repair
• Representation Traceability Restoration
• Identity Boundary Repair
• Synthetic Likeness Governance
• AI Twin Stewardship Review
• Group Truth Pathway Restoration
• Platform Portability Repair
• Appeal Pathway Restoration
• Misrepresentation Repair
• Public Correction
• Canon Summary Repair
• Role Reclarification
• Agency Restoration
• Temporal Revalidation
• Exit Path Creation

Restoration Requirement

Representation failure must repair both the proxy and the represented party’s control over the proxy.

Minimal sequence:

Detect misrepresentation or proxy drift
        ↓
Pause or constrain representation
        ↓
Notify / surface to represented party where appropriate
        ↓
Restore auditability
        ↓
Correct memory / scope / output / authority
        ↓
Repair affected burden
        ↓
Update consent and boundaries
        ↓
Restore rollback and exit
        ↓
Validate over time

10. Domain Expressions

AI Personal Assistants

An AI assistant representing a user must provide:

review
scope control
memory inspection
correction
undo / rollback
action logs
permission boundaries
identity protection
exit

If the assistant sends email, schedules, drafts, negotiates, summarizes, or acts on the user’s behalf, the user must be able to audit the representation continuously.

Convenience does not remove representation rights.

AI Twins / Mirrors

AI twins, mirrors, avatars, synthetic personas, or likeness systems are high-risk representation systems.

They require:

• explicit scope
• ongoing consent
• identity boundary protection
• memory audit
• drift correction
• revocation
• visible distinction where appropriate
• stewardship rules
• transition rules if independent agency ever emerges
• non-consensual modeling protections

A likeness that cannot be controlled by the represented person is not legitimate representation.

AI Governance

AI governance must ensure representation systems include:

contract validity
consent validity
scope clarity
auditability
correction rights
traceability
appeal
rollback
representation boundaries

No AI system should act as representative merely because it can mimic, summarize, or predict.

Representation is a legitimacy claim and must be governed as such.

Security

Representation security includes protection against:

impersonation
credential misuse
voice cloning
avatar abuse
agent takeover
memory poisoning
delegation spoofing
unauthorized signing
identity confusion

AI representation must preserve:

• identity integrity
• authorization trace
• least privilege
• revocation
• audit logs
• recovery pathway

Representation without security becomes impersonation infrastructure.

Governance / JGL

AI may represent citizens, clients, legal parties, employees, institutions, or publics.

Governance requires:

• authority trace
• informed consent
• scope limits
• appeal
• correction
• records
• accountability
• conflict checks
• affected-party protection

AI representation in legal or administrative contexts is invalid if the represented party cannot inspect and correct the proxy.

Economy

AI representation appears in:

customer service
sales agents
hiring profiles
credit profiles
worker scoring
brand voice
contract negotiation
creator avatars
platform personas
automated vendor agents

Economic representation requires:

• clear agency
• disclosure where relevant
• correction
• contract scope
• liability trace
• revocation
• record audit
• anti-impersonation protection

A platform should not own or distort the economic identity of a worker, seller, creator, or customer.

Biology / Medicine

AI may represent patient state through summaries, risk scores, diagnostic models, triage outputs, or medical records.

Patient representation requires:

• patient-accessible correction
• clinician accountability
• evidence trace
• temporal updating
• context preservation
• symptom / lived-response integration
• scope boundaries
• repair after misclassification

A patient’s AI-mediated record must not become more authoritative than the patient’s living system without correction pathway.

CMS / Meaning

AI can represent symbolic meaning, identity, narrative, spiritual orientation, archetype, or inner pattern.

This is high-risk because meaning representation can bind identity.

Requirements:

• provisional framing
• non-identity-binding language
• user correction
• symbolic humility
• context preservation
• time validation
• consent scope
• boundary integrity

AI may mirror meaning.

It must not own meaning.

Principles / Archetypes

AI-generated archetype or principle profiles represent a person or system through symbolic compression.

They must remain:

auditable
provisional
correctable
non-coercive
non-rank-binding
contextual
time-validated

Violation occurs when:

AI names an archetype
        ↓
system treats name as identity
        ↓
µᵢ narrows
        ↓
H↑

Archetype representation must remain user-governed.

Relationships / Couplings

AI representation can enter relationships by drafting messages, summarizing conflict, predicting intent, or advising action.

A relationship becomes incoherent if:

• AI misrepresents one party
• AI drafts without review
• AI amplifies one perspective as truth
• AI stores relational context incorrectly
• AI becomes hidden mediator
• AI simulates consent or intent

Relational AI must preserve the represented party’s agency and correction.

Project / Knowledge Systems

AI representation of UTS canon requires auditability to the project.

AI may summarize, generate, crosswalk, or transform UTS material, but the project must retain:

canon review
version control
definition correction
operator mapping
state-vector verification
cross-link repair
template consistency
deprecation pathways

A generated summary is not canon unless audited and accepted.

AI can represent the archive only under continuous project audit.

11. Scaling Behavior

As AI representation scales, representation risk scales.

Scale increases:

number of represented parties
memory depth
public visibility
automation speed
contractual consequence
identity consequence
reputation risk
correction burden
misrepresentation spread
platform dependency

Therefore:

Representation scale↑ ⇒ audit, correction, consent, and rollback capacity↑

Scaling Risk Pattern

AI representation scale↑
audit capacity flat
correction capacity flat
scope drift↑
misrepresentation H↑

Valid Scaling Pattern

representation scale↑
continuous audit↑
memory correction↑
scope control↑
revocation↑
rollback↑
represented-party sovereignty↑

High-Risk Representation Contexts

High-risk AI representation includes:

• legal representation
• medical representation
• employment representation
• financial representation
• educational records
• synthetic likeness
• AI twins
• public figures
• minors or dependent parties
• institutional authority
• project canon
• group identity
• relationship mediation
• political or civic representation

Relation to INV-065

INV-065 states:

AI is a Γ-amplifier, not a coherence source.

INV-066 specifies:

When AI selects or generates on behalf of a party, the represented party must retain continuous auditability.

Together:

AI can assist representation, but cannot replace represented-party sovereignty.

12. Canonical Examples

Example 1 — AI Email Assistant

An AI writes and sends messages for a user.

If the user cannot review, undo, inspect context, or constrain tone and scope:

proxy action↑
user Au↓
representation risk↑

The assistant became unaccountable representation.

Example 2 — AI Memory Misstates User Preference

AI remembers a user’s preference incorrectly and acts on it repeatedly.

memory drift↑
representation drift↑
µᵢ↓

Memory correction is required.

Example 3 — AI Avatar of a Person

A synthetic avatar speaks as a person after consent becomes outdated or scope expands.

likeness use↑
consent drift↑
identity debt↑

Ongoing consent and auditability are required.

Example 4 — AI Legal Summary Represents Client Position

AI summarizes a client’s legal position incorrectly and the summary circulates internally.

proxy summary↑
client meaning↓
H↑

Representation requires review and correction.

Example 5 — AI Medical Record Summary

AI summarizes a patient record and omits key lived symptoms or recurrence patterns.

summary efficiency↑
organism truth↓
medical risk↑

Patient and clinician correction pathways are required.

Example 6 — AI Represents a Group

AI claims to summarize the views of a community based on incomplete data.

group representation↑
affected-node truth↓
legitimacy debt↑

Group representation requires group truth pathways.

Example 7 — UTS Canon Summary

AI generates a UTS primer that compresses a core concept incorrectly.

If the primer is reused without audit:

summary Φ↑
canon drift↑
archive H↑

Project audit is required before canon use.

13. Anti-Patterns

Anti-Pattern 1 — “It Sounds Like Them”

Likeness is not authorization.

Anti-Pattern 2 — “The AI Knows Their Preferences”

Memory is not legitimate unless inspectable and correctable.

Anti-Pattern 3 — “They Consented Once”

Representation consent must remain scoped and updateable.

Anti-Pattern 4 — “The AI Is More Consistent Than the Person”

Consistency can become proxy capture.

Anti-Pattern 5 — “The Summary Is Good Enough”

A summary representing a party must be correctable by that party.

Anti-Pattern 6 — “It Is Just Automation”

Automation can become agency.

Anti-Pattern 7 — “The Group Can Be Modeled From Data”

Group representation requires affected-node truth, not only data inference.

Anti-Pattern 8 — “Platform Persona Is the User”

The platform’s model of a person is not the person.

Anti-Pattern 9 — “Representation Can Be Corrected Later”

Misrepresentation can spread and create durable debt.

Anti-Pattern 10 — “AI Can Preserve the Canon Automatically”

AI can assist canon. It cannot replace canon audit.

14. Related Laws

This invariant connects strongly to:

• AI Representation Audit Law
• Proxy Capture Law
• Memory Corrigibility Law
• Consent Drift Law
• Representation Boundary Law
• AI Authority Substitution Law
• Storage Is Not Memory Law
• Public Cognition Capture Law
• Affected-Node Truth Law
• Auditability Precedes Legitimacy Law
• Identity Boundary Law
• Synthetic Likeness Governance Law
• Correction Rights Law
• Platform Dependency Law
• Time Validates Law

15. Related Scaling Rules

Related scaling rules:

• Representation Audit Must Scale With Proxy Reach
• Memory Correction Must Scale With Memory Depth
• Consent Renewal Must Scale With Representation Duration
• Rollback Must Scale With Proxy Action Authority
• Traceability Must Scale With Representation Consequence
• Group Truth Pathways Must Scale With Group Representation Claims
• Likeness Governance Must Scale With Synthetic Fidelity
• Platform Portability Must Scale With Identity Dependency
• Correction Capacity Must Scale With Misrepresentation Spread
• AI Agent Permissions Must Scale With Representation Scope
• Canon Review Must Scale With AI-Generated Summaries
• When Audit Cannot Scale, Representation Scope Must Shrink

16. Related Gates

Relevant gates:

• AI Representation Gate
• Representation Auditability Gate
• Consent Validity Gate
• Scope Gate
• Memory Integrity Gate
• Correction Rights Gate
• Rollback Gate
• Traceability Gate
• Identity Integrity Gate
• Synthetic Likeness Gate
• Group Representation Gate
• Platform Portability Gate
• Agent Authority Gate
• Contract Validity Gate
• Public-Impact Gate
• Affected-Node Truth Gate
• Restoration Capacity Gate
• High Risk Gate
• Temporal Validation Gate
• Canon Review Gate

Gate Logic

An AI representation fails the gate when:

represented party cannot audit the proxy

or when:

scope is unclear or expands without consent

or when:

memory cannot be inspected or corrected

or when:

AI acts without traceability

or when:

correction, revocation, or rollback is unavailable

or when:

synthetic likeness continues without ongoing control

or when:

group representation lacks group truth pathways

Gate failure returns:

∅

Meaning:

AI representation is not currently admissible

The coherent response may be:

pause representation
restore auditability
clarify scope
renew consent
repair memory
add correction rights
add rollback
restore traceability
reduce representation authority
validate over time

17. Related Operators

18. Machine-Readable Summary

id: UTS-INV-066
name: AI Representation Requires Continuous Auditability to the Represented Party
registry: UTS Invariants Registry
category: AI Governance Invariant / Representation Invariant / Auditability Invariant / Boundary Invariant
status: Draft-Integrated
version: 0.1

definition: >
  A representation system that cannot be audited by the represented party is
  not legitimate representation. AI representation occurs when an AI system
  acts for, speaks for, models, mirrors, summarizes, simulates, negotiates for,
  remembers for, decides for, recommends for, ranks for, or presents a person,
  group, institution, community, role, identity, brand, archive, or public.

constraint: >
  Any AI system that represents a person, group, institution, role, identity,
  archive, public, or meaning structure must remain continuously auditable,
  correctable, scoped, revocable, and appealable by the represented party,
  with clear boundaries, traceability, rollback, and restoration pathways.

canonical_form:
  - "AI representation requires continuous auditability to the represented party"
  - "No representation without audit"
  - "The represented party must be able to inspect and correct the representation"
  - "Representation is not legitimate unless the represented party remains sovereign over the proxy"
  - "Likeness is not authorization"
  - "Platform model of a person is not the person"
  - "AI can assist representation, but cannot replace represented-party sovereignty"

protects:
  - represented_party_sovereignty
  - auditability
  - meaning_integrity
  - identity_integrity
  - boundary_integrity
  - consent_validity
  - memory_integrity
  - correction_rights
  - rollback_capacity
  - representation_legitimacy

state_vector_effects_when_preserved:
  O: "stable_or_increasing_because_representation_remains_accountable"
  H: "contained_because_proxy_drift_routes_to_correction"
  ε: "visible_misrepresentation_is_correctable_and_repairable"
  ι: "decreases_because_proxy_is_not_misread_as_represented_party"
  Au: "continuous_to_the_represented_party"
  µᵢ: "preserved_through_meaning_identity_and_context_integrity"
  BΣ: "preserved_through_scope_consent_and_boundary_control"
  K: "maintained_between_proxy_and_represented_party"
  R: "available_for_misrepresentation_memory_and_identity_repair"
  Φ: "likeness_accuracy_convenience_automation_or_engagement_not_misread_as_legitimacy"

state_vector_effects_when_violated:
  O: "decreases_as_proxy_capture_or_drift_accumulates"
  H: "increases_through_unrepaired_misrepresentation"
  ε: "appears_as_misrepresentation_impersonation_drift_mistaken_action_or_trust_failure"
  ι: "increases_when_representation_is_misread_as_source_authority"
  Au: "decreases_when_represented_party_cannot_inspect_or_correct"
  µᵢ: "degrades_when_identity_meaning_or_context_is_distorted"
  BΣ: "decreases_when_scope_consent_or_proxy_boundary_fails"
  K: "declines_between_AI_proxy_and_represented_party"
  R: "insufficient_if_correction_revocation_or_rollback_is_unavailable"
  Φ: "may_rise_through_usefulness_likeness_productivity_or_automation"

primary_u_layer: U2
memory_layer: U7
classification_layer: U4
execution_layer: U3
field_layer: U6
coordination_layer: U5
resource_layer: U1
environment_layer: U8

violation_signatures:
  - ai_speaks_for_a_party_without_review
  - memory_drift_becomes_representation_drift
  - scope_creep_in_delegation
  - proxy_acts_without_traceability
  - represented_party_cannot_correct_the_record
  - synthetic_likeness_without_ongoing_control
  - group_representation_without_group_truth_pathways
  - ai_representation_used_as_authority
  - platform_owns_the_representation_layer
  - project_canon_misrepresented_by_ai_summary

related_failure_modes:
  - AI Representation Capture
  - Proxy Drift
  - Memory Based Misrepresentation
  - Scope Creep In Delegation
  - Unauthorized Agency
  - Synthetic Likeness Capture
  - Persona Drift
  - AI Twin Capture
  - Group Representation Without Truth
  - Proxy Authority Inversion
  - Representation Without Auditability
  - Correction Path Failure
  - Consent Drift
  - Memory Non Corrigibility
  - Identity Boundary Failure
  - Platform Representation Capture
  - Impersonation Risk
  - Public Cognition Misrepresentation
  - Canon Misrepresentation
  - Meaning Integrity Loss
  - Hidden Debt Accumulation
  - Legitimacy Debt
  - Restoration Capacity Lag

related_restoration_arcs:
  - Representation Audit Restoration
  - Memory Correction
  - Scope Reconstitution
  - Consent Renewal
  - Proxy Authority Rollback
  - Correction Pathway Repair
  - Representation Traceability Restoration
  - Identity Boundary Repair
  - Synthetic Likeness Governance
  - AI Twin Stewardship Review
  - Group Truth Pathway Restoration
  - Platform Portability Repair
  - Appeal Pathway Restoration
  - Misrepresentation Repair
  - Public Correction
  - Canon Summary Repair
  - Role Reclarification
  - Agency Restoration
  - Temporal Revalidation
  - Exit Path Creation

related_laws:
  - AI Representation Audit Law
  - Proxy Capture Law
  - Memory Corrigibility Law
  - Consent Drift Law
  - Representation Boundary Law
  - AI Authority Substitution Law
  - Storage Is Not Memory Law
  - Public Cognition Capture Law
  - Affected Node Truth Law
  - Auditability Precedes Legitimacy Law
  - Identity Boundary Law
  - Synthetic Likeness Governance Law
  - Correction Rights Law
  - Platform Dependency Law
  - Time Validates Law

related_scaling_rules:
  - Representation Audit Must Scale With Proxy Reach
  - Memory Correction Must Scale With Memory Depth
  - Consent Renewal Must Scale With Representation Duration
  - Rollback Must Scale With Proxy Action Authority
  - Traceability Must Scale With Representation Consequence
  - Group Truth Pathways Must Scale With Group Representation Claims
  - Likeness Governance Must Scale With Synthetic Fidelity
  - Platform Portability Must Scale With Identity Dependency
  - Correction Capacity Must Scale With Misrepresentation Spread
  - AI Agent Permissions Must Scale With Representation Scope
  - Canon Review Must Scale With AI Generated Summaries
  - When Audit Cannot Scale Representation Scope Must Shrink

related_gates:
  - AI Representation Gate
  - Representation Auditability Gate
  - Consent Validity Gate
  - Scope Gate
  - Memory Integrity Gate
  - Correction Rights Gate
  - Rollback Gate
  - Traceability Gate
  - Identity Integrity Gate
  - Synthetic Likeness Gate
  - Group Representation Gate
  - Platform Portability Gate
  - Agent Authority Gate
  - Contract Validity Gate
  - Public Impact Gate
  - Affected Node Truth Gate
  - Restoration Capacity Gate
  - High Risk Gate
  - Temporal Validation Gate
  - Canon Review Gate

19. Compact Canon Statement

UTS-INV-066 states that AI representation requires continuous auditability to the represented party. When AI acts for, speaks for, models, mirrors, remembers for, decides for, or presents a person, group, institution, role, archive, or public, the represented party must be able to inspect, correct, scope, revoke, appeal, and exit the representation. Representation without audit becomes proxy capture. Likeness is not authorization. A platform’s model of a person is not the person. AI can assist representation, but cannot replace represented-party sovereignty.

20. Short Reference Version

UTS-INV-066 — AI Representation Requires Continuous Auditability to the Represented Party

No representation without audit.

AI representation occurs when AI:

speaks for
acts for
models
mirrors
remembers for
decides for
summarizes
negotiates
presents
or simulates

a person, group, institution, archive, role, identity, or public.

The represented party must retain:

inspection
correction
scope control
revocation
appeal
rollback
exit
memory audit
traceability

Violation pattern:

AI proxy active
represented-party Au↓
scope drift↑
memory drift↑
correction unavailable
µᵢ↓
BΣ↓
H↑
ι↑

Core rule:

Likeness is not authorization.
The platform model of a person is not the person.
Representation without audit becomes proxy capture.