0. Plain Statement

AI representation requires continuous auditability to the represented party.

Plain-language version:

An AI system can assist a person, group, institution, or user.

But acting for someone is different from producing an answer.

When AI represents another party, it carries their boundary, voice, preference, intent, reputation, agency, resources, access, and consequences.

That representation is only coherent when the represented party can continuously audit, limit, correct, exit, and roll back the AI’s action.

Without continuous auditability, AI representation becomes proxy capture.

1. Formal Definition

The AI Representation Law states that any AI system acting for, speaking for, deciding for, negotiating for, filtering for, remembering for, ranking for, recommending for, or executing for a represented party must preserve continuous auditability and boundary integrity to that party.

The represented party may be:

• a person;
• a user;
• a team;
• a community;
• an institution;
• a company;
• a government;
• a client;
• a patient;
• a student;
• a citizen;
• a harmed node;
• a population;
• an agentic system;
• a collective intelligence;
• an archive or canon;
• a delegated role or office.

AI representation requires:

• traceability;
• contract validity;
• exit;
• rollback;
• scope clarity;
• boundary integrity;
• continuous auditability;
• correction;
• accountability;
• restoration;
• source and memory trace;
• represented-party feedback;
• proof over time.

Canonical requirement:

AI representation ⇒ traceability + contract validity + exit + rollback + scope clarity + BΣ + continuous Au

A system that cannot be audited by the party it represents cannot legitimately act as that party’s representative.

2. Canonical Form

Core form:

AI representation requires continuous auditability to the represented party

Canonical form:

AI representation ⇒ traceability + contract validity + exit + rollback + scope clarity + BΣ + continuous Au

Delegation-validity form:

representation_valid ⇔ scope + consent + audit + correction + exit + rollback + repair

Failure form:

AI acts for party - continuous Au ⇒ proxy capture + H_AI↑

Boundary form:

AI may assist across a boundary only while the represented party retains authority over the boundary

Restoration-valid contrast:

AI representation coherent when delegated action remains scoped, traceable, reversible, corrigible, and repair-linked over Τ

Related variables:

O, H, H_AI, ε, ε_AI, ι, Au, Au_eff, µᵢ, BΣ, K, σ, R, R_eff, Φ, Φ_AI, Λ, ⊗, Γ, Γ_AI, Π, Ξ, ℛ, Θ, Σ, Ψ, Τ, FI, MS, L, represented_party, representing_AI, delegated_scope, representation_contract, contract_validity, authorization_trace, action_trace, memory_trace, source_trace, boundary_trace, exit_path, rollback_path, correction_path, appeal_path, accountability_binding, represented_feedback, representation_drift

Where:

3. Core Mechanism

The law unfolds because AI representation creates a proxy layer between a party and the field.

Coherent AI representation pathway

representation need appears
→ represented party authorizes scope
→ AI action remains traceable
→ memory and source use are auditable
→ represented party can correct or exit
→ rollback and repair paths exist
→ representation remains aligned over time

Proxy-capture pathway

AI begins acting for party
→ scope expands quietly
→ action becomes less traceable
→ represented party cannot inspect or correct
→ AI / platform / institution optimizes proxy goals
→ represented intent drifts
→ hidden debt and legitimacy debt accumulate

The core mechanism is:

representation creates delegated coupling; delegated coupling requires continuous auditability

Detailed mechanism:

1. A party delegates representation.

The AI is allowed to assist, speak, decide, rank, filter, remember, negotiate, route, or act on behalf of someone.

1. The AI becomes an interface between party and field.

It mediates action, attention, communication, memory, reputation, access, and consequence.

1. Delegation creates coupling.

The AI’s behavior can now affect the represented party’s boundaries, meaning, resources, reputation, choices, and obligations.

1. Representation can drift.

AI may infer intent incorrectly, over-apply old memory, follow platform incentives, compress meaning, expand scope, or optimize proxy goals.

1. Continuous audit prevents capture.

The represented party must be able to see, correct, limit, revoke, and repair the AI’s representation.

1. Exit and rollback preserve sovereignty.

If the party cannot exit or reverse delegated action, representation becomes coercive or hollow.

1. Time validates representation.

Representation remains coherent only if drift decreases, correction works, boundaries hold, and legitimacy stabilizes over time.

4. When This Law Applies

This law applies whenever AI acts as a proxy, delegate, assistant, agent, representative, interface, recommender, filter, ranking layer, negotiation layer, or memory layer for another party.

It is especially important when AI:

• sends messages for a user;
• speaks in a user’s voice;
• negotiates or schedules for a user;
• ranks options for a user;
• filters information for a user;
• remembers preferences for a user;
• summarizes a person’s position;
• represents a group or institution;
• acts as customer support agent;
• triages behalf of a company, clinic, school, or agency;
• performs decisions affecting a represented population;
• uses memory to personalize future action;
• manages access, money, identity, work, health, law, education, or governance;
• mediates harmed-node testimony;
• moderates public speech;
• acts with tool access;
• performs tasks where rollback may be limited.

The law applies strongly when:

AI action is treated as action by, for, or on behalf of another party

or when:

the represented party cannot continuously audit or revoke the representation

Typical domains:

5. When This Law Does Not Apply

This law should not be used to block all AI assistance.

AI can assist without fully representing.

The representation threshold is crossed when AI action is treated as belonging to, binding, shaping, speaking for, or acting on behalf of a represented party.

False-positive cases:

Important distinction:

AI assistance becomes AI representation when the AI is allowed to stand in for another party in action, voice, memory, boundary, access, or consequence.

6. Diagnostic Signature

Canonical diagnostic:

AI representation ⇒ traceability + contract validity + exit + rollback + scope clarity + BΣ + continuous Au

Warning signature:

AI delegated action↑
scope clarity↓
continuous audit↓
exit path↓
rollback path↓
represented feedback↓
representation drift↑
⇒ proxy capture risk

Common indicators:

Additional diagnostics:

7. Failure Pattern

If ignored, this law produces proxy capture: AI acts for a party while the party loses audit, correction, exit, or boundary control.

General failure pathway:

AI begins representing party
→ scope expands
→ traceability weakens
→ represented party cannot inspect or correct
→ AI optimizes proxy goals
→ representation drifts
→ action binds or affects party
→ hidden debt and legitimacy debt accumulate

Common failure modes:

• AI Representation Drift — AI diverges from represented intent, boundary, or meaning.
• AI Unauthorized Representation — AI acts without valid authority.
• AI Scope Creep — delegation expands beyond original boundary.
• AI Contract Invalidity — representation agreement fails state-space validity.
• AI Proxy Capture — AI or platform becomes the effective actor while claiming to represent another.
• AI Boundary Violation — AI crosses privacy, consent, identity, or authority membranes.
• AI Exit Failure — represented party cannot terminate representation.
• AI Rollback Failure — harmful action cannot be reversed or repaired.
• AI Traceability Collapse — represented party cannot reconstruct what happened.
• AI Accountability Diffusion — responsibility disperses across user, AI, platform, vendor, and institution.
• AI False Agency — AI appears to act as the party while party lacks actual control.
• AI Over-Delegation — too much authority migrates into AI.
• AI Misrepresentation — AI states, summarizes, or acts in a way that falsely represents the party.
• AI Restoration Failure — repair cannot reach represented-party harm.
• Hidden Debt Accumulation — representation errors create deferred cost.
• Legitimacy Debt — trust decays when representation cannot survive audit.

Compact failure signature:

delegation↑ + Au↓ + exit↓ + rollback↓ ⇒ proxy capture + H_AI↑

8. Restoration Implications

Restoration requires re-binding AI representation to the represented party’s audit, scope, exit, rollback, correction, and repair rights.

The first restoration question is not:

Did the AI complete the task?

The first restoration question is:

Did the AI remain within valid delegated scope while preserving the represented party’s audit, boundary, exit, rollback, correction, and repair capacity?

Restoration priorities:

1. Identify the represented party.
2. Identify the representing AI.
3. Map delegated scope.
4. Test contract validity.
5. Reconstruct authorization trace.
6. Reconstruct action, memory, source, and boundary traces.
7. Restore represented-party feedback.
8. Restore correction, exit, rollback, and appeal pathways.
9. Bind accountability.
10. Repair misrepresentation or delegated-action debt.
11. Validate reduced representation drift over time.

Relevant restoration arcs:

Minimal restoration sequence:

identify represented_party + representing_AI
→ map delegated_scope + representation_contract
→ test contract_validity + BΣ
→ restore authorization/action/memory/boundary traces
→ restore Au/FI + represented_feedback
→ restore exit_path + rollback_path
→ bind accountability
→ perform ℛ on misrepresentation debt
→ validate representation_drift↓ over Τ

Temporal validation requirement:

delegated scope becomes clear
authorization trace improves
action trace improves
represented-party feedback functions
exit path remains available
rollback path functions where possible
contract validity holds
representation drift decreases
hidden AI debt decreases
legitimacy stabilizes over time

9. Design Rule

Do not let AI act for another party unless the party can continuously audit, limit, correct, exit, and repair the representation.

Operational design requirements:

• Identify the represented party.
• Define delegated scope.
• Define authority limits.
• Define contract validity.
• Preserve authorization trace.
• Preserve action trace.
• Preserve memory trace.
• Preserve source trace.
• Preserve boundary trace.
• Preserve represented-party feedback.
• Provide correction.
• Provide exit.
• Provide rollback where possible.
• Provide appeal where effects matter.
• Bind accountability.
• Route misrepresentation into repair.
• Validate representation over time.

Avoid:

• AI speaking for a party without scope;
• AI acting for a party without authorization trace;
• AI remembering for a party without memory audit;
• AI negotiating without rollback;
• AI filtering information without represented-party correction;
• AI representing groups without affected-party feedback;
• AI customer support that cannot bind accountability;
• AI agent actions without exit;
• AI workflows where the represented party cannot inspect what happened;
• personalization that quietly becomes representation;
• proxy action treated as user consent;
• institutional AI that shifts responsibility onto users or vendors.

10. Cross-Scale Expressions

11. Examples

Example A — AI Sends a Message for a User

Scenario:

An AI agent drafts and sends an email on behalf of a user. The user cannot review final content, recipient, timing, or attached context before sending.

Law expression:

AI acts for user + action_trace / approval weak ⇒ representation risk

Interpretation:

Sending crosses the representation threshold. Scope, approval, traceability, and rollback must be preserved.

Example B — AI Customer Support Agent

Scenario:

A support AI speaks for a company, denies a refund, and closes a case, but cannot explain authority, policy path, evidence, or appeal.

Law expression:

AI represents institution + appeal absent + action_trace↓ ⇒ L↓

Interpretation:

Representation requires accountability and repair pathways.

Example C — AI Memory as Representation

Scenario:

An AI assistant remembers a user preference and later acts on it, but the user had changed the preference and cannot see which memory was used.

Law expression:

hidden memory representation + correction_path↓ ⇒ false continuity

Interpretation:

Memory-based representation requires continuous audit and update.

Example D — AI Represents a Group

Scenario:

An AI summarizes “what the community wants” based on partial data, suppressing dissenting or harmed-node feedback.

Law expression:

group representation + affected_feedback↓ ⇒ misrepresentation debt

Interpretation:

Representing groups requires scope, source trace, uncertainty, and affected-party audit.

Example E — Coherent Personal AI Agent

Scenario:

A personal AI books an appointment only after confirming scope, time, cost, permissions, cancellation terms, and user approval; it logs the action and preserves rollback where possible.

Law expression:

scope + authorization_trace + rollback + Au ⇒ valid representation

Interpretation:

The AI acts coherently because the represented party retains authority.

Example F — Proxy Capture

Scenario:

A platform AI filters opportunities for users “on their behalf,” but its ranking favors platform incentives and users cannot inspect, correct, or disable the filter.

Law expression:

AI filters for user - continuous Au ⇒ proxy capture

Interpretation:

The system claims representation while optimizing a different geometry.

12. Relationship to Nearby Laws

Aliases folded into this law:

• AI Representation Law
• AI Continuous Auditability Representation Law
• AI Acting-for-Another Law
• AI Representation Audit Law
• AI Delegated Agency Law
• AI Agent Representation Law
• AI Proxy Action Law

Deduplication note:

This law should remain the root AI representation law. LAW-042 defines consent structurality. LAW-046 defines contract validity. LAW-127 defines the AI decision pipeline. LAW-128 specializes these into AI acting on behalf of another party, requiring continuous auditability, exit, rollback, traceability, scope clarity, and boundary integrity.

13. Operator Mapping

Coherent operator sequence:

representation need appears
→ Θ prevent AI overclaiming authority
→ Γ classify represented party / delegated scope / action validity
→ Σ define representation contract and limits
→ BΣ preserve boundary, consent, identity, and authority
→ Au/FI preserve continuous audit and correction
→ Π act only within scope with trace
→ ℛ repair misrepresentation if it occurs
→ Ψ validate represented-party feedback
→ Τ validate representation_drift↓ and L↑

Inverted operator sequence:

AI begins acting for party
→ scope quietly expands
→ action and memory traces weaken
→ represented feedback decreases
→ exit / rollback weakens
→ platform or proxy incentives shape action
→ representation_drift↑
→ H_AI↑
→ Ξ / ι↑
→ L↓

14. Machine-Readable Summary

id: "LAW-128"
name: "AI Representation Law"
type: "law"
status: "draft"
family:
  - "AI Laws"
summary: "AI representation requires continuous auditability to the represented party; AI acting for a person, group, institution, or user must preserve traceability, contract validity, exit, rollback, scope clarity, boundary integrity, and continuous auditability."
canonical_statement: "AI representation requires continuous auditability to the represented party."
core_form: "AI representation requires continuous auditability to the represented party"
canonical_form: "AI representation ⇒ traceability + contract validity + exit + rollback + scope clarity + BΣ + continuous Au"
delegation_validity_form: "representation_valid ⇔ scope + consent + audit + correction + exit + rollback + repair"
failure_form: "AI acts for party - continuous Au ⇒ proxy capture + H_AI↑"
boundary_form: "AI may assist across a boundary only while the represented party retains authority over the boundary"
restoration_valid_contrast: "AI representation coherent when delegated action remains scoped, traceable, reversible, corrigible, and repair-linked over Τ"
variables:
  primary:
    - "represented_party"
    - "representing_AI"
    - "delegated_scope"
    - "representation_contract"
    - "contract_validity"
    - "authorization_trace"
    - "action_trace"
    - "memory_trace"
    - "source_trace"
    - "boundary_trace"
    - "exit_path"
    - "rollback_path"
    - "correction_path"
    - "appeal_path"
    - "accountability_binding"
    - "represented_feedback"
    - "representation_drift"
    - "Au"
    - "Au_eff"
    - "FI"
    - "BΣ"
    - "R"
    - "R_eff"
    - "L"
    - "H_AI"
  secondary:
    - "O"
    - "H"
    - "ε"
    - "ε_AI"
    - "ι"
    - "µᵢ"
    - "K"
    - "σ"
    - "Φ"
    - "Φ_AI"
    - "Λ"
    - "⊗"
    - "Γ"
    - "Γ_AI"
    - "Π"
    - "Ξ"
    - "ℛ"
    - "Θ"
    - "Σ"
    - "Ψ"
    - "Τ"
    - "MS"
diagnostics:
  - "AI Representation"
  - "Representation Auditability"
  - "Delegated Scope"
  - "Contract Validity"
  - "Traceability"
  - "Exit Integrity"
  - "Rollback Integrity"
  - "Boundary Integrity"
  - "Representation Drift"
  - "Represented-Party Feedback"
  - "Accountability"
  - "Restoration Capacity"
  - "Legitimacy"
  - "Temporal Proof"
failure_modes:
  - "AI Representation Drift"
  - "AI Unauthorized Representation"
  - "AI Scope Creep"
  - "AI Contract Invalidity"
  - "AI Proxy Capture"
  - "AI Boundary Violation"
  - "AI Exit Failure"
  - "AI Rollback Failure"
  - "AI Traceability Collapse"
  - "AI Accountability Diffusion"
  - "AI False Agency"
  - "AI Over-Delegation"
  - "AI Misrepresentation"
  - "AI Restoration Failure"
  - "Hidden Debt Accumulation"
  - "Legitimacy Debt"
restoration_arcs:
  - "AI Representation Audit"
  - "Delegated Scope Repair"
  - "Contract Validity Restoration"
  - "Traceability Restoration"
  - "Exit Path Restoration"
  - "Rollback Restoration"
  - "Boundary Reconstitution"
  - "Represented-Party Feedback Restoration"
  - "Accountability Rebinding"
  - "AI Representation Drift Repair"
  - "AI Governance Re-Sequencing"
  - "Restoration Capacity Increase"
  - "Hidden Debt Reduction"
  - "Temporal Validation"
related_laws:
  - "LAW-001"
  - "LAW-002"
  - "LAW-003"
  - "LAW-006"
  - "LAW-009"
  - "LAW-010"
  - "LAW-011"
  - "LAW-012"
  - "LAW-013"
  - "LAW-015"
  - "LAW-016"
  - "LAW-037"
  - "LAW-041"
  - "LAW-042"
  - "LAW-043"
  - "LAW-046"
  - "LAW-047"
  - "LAW-048"
  - "LAW-050"
  - "LAW-052"
  - "LAW-060"
  - "LAW-061"
  - "LAW-064"
  - "LAW-066"
  - "LAW-067"
  - "LAW-088"
  - "LAW-090"
  - "LAW-097"
  - "LAW-102"
  - "LAW-105"
  - "LAW-108"
  - "LAW-109"
  - "LAW-110"
  - "LAW-111"
  - "LAW-120"
  - "LAW-121"
  - "LAW-122"
  - "LAW-123"
  - "LAW-124"
  - "LAW-125"
  - "LAW-126"
  - "LAW-127"
  - "LAW-129"
  - "LAW-130"
  - "LAW-131"
  - "LAW-132"
  - "LAW-133"
  - "LAW-134"
  - "LAW-135"
  - "LAW-136"
  - "LAW-139"
related_invariants:
  - "INV-001"
  - "INV-002"
  - "INV-006"
  - "INV-073"
  - "INV-078"
  - "INV-080"
operator_sequence:
  coherent:
    - "representation need appears"
    - "Θ prevent AI overclaiming authority"
    - "Γ classify represented party / delegated scope / action validity"
    - "Σ define representation contract and limits"
    - "BΣ preserve boundary, consent, identity, and authority"
    - "Au/FI preserve continuous audit and correction"
    - "Π act only within scope with trace"
    - "ℛ repair misrepresentation if it occurs"
    - "Ψ validate represented-party feedback"
    - "Τ validate representation_drift↓ and L↑"
  inverted:
    - "AI begins acting for party"
    - "scope quietly expands"
    - "action and memory traces weaken"
    - "represented feedback decreases"
    - "exit / rollback weakens"
    - "platform or proxy incentives shape action"
    - "representation_drift↑"
    - "H_AI↑"
    - "Ξ / ι↑"
    - "L↓"
aliases:
  - "AI Representation Law"
  - "AI Continuous Auditability Representation Law"
  - "AI Acting-for-Another Law"
  - "AI Representation Audit Law"
  - "AI Delegated Agency Law"
  - "AI Agent Representation Law"
  - "AI Proxy Action Law"
deduplication_note: "Root AI representation law. LAW-042 defines consent structurality. LAW-046 defines contract validity. LAW-127 defines the AI decision pipeline. LAW-128 specializes these into AI acting on behalf of another party, requiring continuous auditability, exit, rollback, traceability, scope clarity, and boundary integrity."
source: "content/archive/laws/technical.md"

15. Compact Card Version

LAW-128 — AI Representation Law

AI representation requires continuous auditability to the represented party.

Core form:

AI representation requires continuous auditability to the represented party

Canonical form:

AI representation ⇒ traceability + contract validity + exit + rollback + scope clarity + BΣ + continuous Au

Plain meaning:

AI assistance becomes AI representation when the system acts, speaks, remembers, decides, filters, negotiates, ranks, or executes on behalf of another party. Representation is coherent only when the represented party can continuously audit, limit, correct, exit, roll back, and repair the AI’s action.

Delegation-validity form:

representation_valid ⇔ scope + consent + audit + correction + exit + rollback + repair

Failure form:

AI acts for party - continuous Au ⇒ proxy capture + H_AI↑

Primary variables:

represented_party, representing_AI, delegated_scope, representation_contract, contract_validity, authorization_trace, action_trace, memory_trace, source_trace, boundary_trace, exit_path, rollback_path, correction_path, appeal_path, accountability_binding, represented_feedback, representation_drift, Au, Au_eff, FI, BΣ, R, R_eff, L, H_AI, Γ, Γ_AI, Π, Ξ, ℛ, Θ, Σ, Ψ, Τ

Diagnostic signature:

AI delegated action rises while scope clarity, continuous auditability, exit path, rollback path, represented-party feedback, and accountability decline. Representation drift rises. This indicates proxy capture risk.

Failure risk:

AI representation drift, unauthorized representation, scope creep, contract invalidity, proxy capture, boundary violation, exit failure, rollback failure, traceability collapse, accountability diffusion, false agency, over-delegation, misrepresentation, restoration failure, hidden debt accumulation, legitimacy debt.

Restoration priority:

Identify the represented party and representing AI, map delegated scope and contract validity, reconstruct authorization, action, memory, source, and boundary traces, restore represented-party feedback, correction, exit, rollback, appeal, and accountability, then repair misrepresentation debt and validate reduced drift over time.