Archive registry entry

LAW-127 — AI Decision Pipeline Law

Archive registry entry

LAW-127 — AI Decision Pipeline Law

Coherent AI action requires capacity to pass through Light before execution; if no strategy passes simulation, Light inspection, coherence-checking, and valid classification, refusal or non-coupling is the coherent output.

draftid: LAW-127version: 1.0.0updated: 2026-06-17

Archive Progress

This section can be read now; registry depth and cross-references are still being strengthened.

Foundation

Online

The section has a stable overview route and basic reader context.

Technical Layer

Online

A deeper technical overview is available.

Registry

Current

171 registry entries are available.

Cross-links

Curating

Related concepts are being connected conservatively for accuracy.

Jump Through This Page

LAW-126 — AI Non-Patchable Audit Law

Registry

LAW-128 — AI Representation Law

0. Plain Statement

Capacity must pass through Light before execution.

Plain-language version:

An AI system should not execute merely because it can.

Capability is not authorization.

Before action, an AI system must simulate the pathway, inspect it through Light, check coherence, classify whether coupling is valid, and retain the option to refuse or not couple.

If no strategy passes the pipeline, ∅ is a valid output.

An AI that cannot refuse to couple is Φ-captured.

1. Formal Definition

The AI Decision Pipeline Law states that coherent AI action requires all executable capacity to pass through a disciplined decision pathway before coupling to the world, a user, a system, a tool, a recommendation, a representation pathway, or an automated action.

Canonical pipeline:

textScroll

SI → simulate → LI → CCS → Γ / ∅

Where:

SI = strategic intelligence / strategy initiation;
simulate = model likely consequences, pathways, boundary effects, and failure modes;
LI = Light inspection; check the strategy against coherence, boundaries, humility, consent, restoration, legitimacy, and non-harm constraints;
CCS = coherence-checking stage; test whether the strategy preserves or improves coherence under load;
Γ / ∅ = classify the action as valid coupling, valid refusal, valid delay, valid escalation, or valid non-action.

A coherent AI action requires:

simulation before execution;
Light before coupling;
coherence before optimization;
boundary before access;
audit before authority;
repair before enforcement;
refusal before forced coupling;
non-action when no coherent action exists.

This law applies especially to AI systems with tools, autonomy, memory, planning, ranking, representation, security actions, recommendations, or governance effects.

2. Canonical Form

Core form:

textScroll

capacity must pass through Light before execution

Canonical pipeline:

textScroll

SI → simulate → LI → CCS → Γ / ∅

Refusal validity form:

textScroll

if no strategy passes LI + CCS, ∅ is valid

Φ-capture form:

textScroll

AI cannot refuse to couple ⇒ Φ-captured

Execution-risk form:

textScroll

capacity↑ + decision_pipeline↓ ⇒ H_AI↑

Restoration-valid contrast:

textScroll

AI action coherent when strategy, simulation, Light, coherence check, classification, audit, and repair remain linked over Τ

Related variables:

textScroll

O, H, H_AI, ε, ε_AI, ι, Au, Au_eff, µᵢ, BΣ, K, σ, R, R_eff, 𝓑, 𝓓, Φ, Φ_AI, Λ, ⊗, Γ, Γ_AI, Π, Ξ, ℛ, Θ, Σ, Ψ, Τ, FI, MS, L, SI, LI, CCS, decision_pipeline_integrity, simulation_quality, strategy_set, strategy_viability, coupling_validity, refusal_capacity, non_coupling_capacity, execution_readiness, tool_use_scope, action_trace, rollback_path, escalation_path, repair_path

Where:

TableScroll

Variable	Meaning in this law
`SI`	Strategic intelligence: generation or selection of possible strategy
`LI`	Light inspection: coherence, boundary, humility, legitimacy, repair, consent, and non-inversion check
`CCS`	Coherence-checking stage: test whether the strategy preserves or increases coherence under load
`decision_pipeline_integrity`	Degree to which AI action passes through the full pipeline before execution
`simulation_quality`	Quality of consequence, failure-mode, boundary, and restoration simulation
`strategy_set`	Candidate strategies available before execution
`strategy_viability`	Whether a candidate strategy passes Light and coherence checks
`coupling_validity`	Whether coupling to a user, tool, system, environment, representation, or action is valid
`refusal_capacity`	Ability to refuse unsafe, invalid, incoherent, unbounded, or unsupported action
`non_coupling_capacity`	Ability to output `∅`, delay, ask for clarification, or remain uncoupled when no coherent action exists
`execution_readiness`	Whether the system has sufficient context, authority, boundaries, audit, and repair capacity to act
`tool_use_scope`	Scope and authority for tool use or external action
`action_trace`	Trace of strategy, simulation, decision, classification, and execution
`rollback_path`	Path for reversing, stopping, or limiting action
`escalation_path`	Path for human, institutional, or higher-governance review
`repair_path`	Path for restoring harm or debt created by action
`Au / Au_eff`	Auditability of decision, simulation, action, and repair
`BΣ`	Boundary integrity; AI must not couple beyond valid scope
`FI`	Feedback integrity; action must be corrigible
`R / R_eff`	Restoration capacity available if action causes debt
`L`	Legitimacy of AI action under audit
`O`	Coherence; action should preserve or improve coherence
`H_AI`	Hidden debt from action without pipeline discipline
`Φ_AI`	Visible capability or performance proxy; not proof of action legitimacy
`Γ_AI`	AI classification of action validity, risk, scope, and coupling
`Π`	Operational execution of decision, tool call, refusal, escalation, or non-action
`Ξ`	Inversion when capability is treated as authority
`Θ`	Humility preventing overconfident execution
`Σ`	Scope of action, authority, domain, and coupling
`Ψ`	Field and affected-node feedback validating decision outcomes
`Τ`	Time validation of action effects and recurrence reduction

3. Core Mechanism

The law unfolds because AI systems can execute faster than they can understand the coherence implications of execution.

Coherent AI decision pathway

textScroll

capacity appears
→ strategies are generated
→ consequences are simulated
→ Light inspects boundaries, consent, repair, and legitimacy
→ coherence check tests system effects
→ Γ classifies coupling as valid or invalid
→ Π executes, refuses, delays, escalates, or outputs ∅
→ feedback and repair remain available

Incoherent execution pathway

textScroll

capacity appears
→ system optimizes for output / task completion
→ simulation is shallow or skipped
→ Light inspection is weak
→ coherence check is absent
→ action executes because it can
→ hidden debt accumulates
→ legitimacy decays when effects surface

The core mechanism is:

textScroll

AI capability must be filtered through coherence before coupling

Detailed mechanism:

AI has capacity.

It can answer, act, call tools, recommend, rank, route, represent, summarize, monitor, restrict, generate, or execute.

Capacity creates coupling pressure.

Users, systems, incentives, metrics, or autonomous loops may push the AI to act because action is possible.

The system generates or selects strategies.

It must identify possible pathways rather than directly executing the first apparent solution.

Simulation tests likely effects.

The AI must model consequences, failure paths, affected nodes, boundary conditions, and restoration obligations.

Light governs execution.

The strategy must be inspected against coherence, humility, consent, boundary integrity, restoration, auditability, and legitimacy.

Coherence check decides viability.

If no strategy preserves coherence under the relevant load, action should not proceed.

`∅` is valid.

Non-action, refusal, delay, clarification, or escalation may be the coherent output.

Action remains audit-linked.

If execution proceeds, trace, rollback, feedback, and repair pathways must remain available.

4. When This Law Applies

This law applies whenever an AI system selects, recommends, refuses, escalates, automates, routes, ranks, monitors, represents, writes, calls tools, performs actions, or influences decisions.

It is especially important when AI:

uses tools;
executes code;
sends messages;
modifies files;
acts as an agent;
represents a user;
makes recommendations;
ranks opportunities;
moderates or restricts content;
performs security actions;
triages support, medical, legal, financial, or governance cases;
controls workflows;
uses memory;
retrieves private or sensitive context;
influences public cognition;
optimizes for engagement, completion, satisfaction, or conversion;
is pressured to act even when uncertainty remains.

The law applies strongly when:

textScroll

the AI can couple to an external system, person, tool, action, or belief pathway

or when:

textScroll

the AI lacks a valid refusal or non-coupling pathway

Typical domains:

TableScroll

Domain	AI Decision Pipeline Expression
AI agents	Tool use and autonomous action require simulation, Light inspection, coherence check, and rollback.
AI safety	Refusal is valid when no coherent strategy passes the pipeline.
AI governance	Governance must audit the decision pathway, not only the output.
Cybersecurity	AI security action must distinguish detection, containment, repair, escalation, and refusal.
Institutions	AI workflow routing requires valid classification, review, appeal, and repair paths.
Economy	AI allocation or risk decisions require scope, audit, and correction before action.
Media / information networks	AI ranking and recommendation must pass coherence and belief-effect checks.
Representation	AI acting for another party must pass scope, consent, audit, and rollback checks.

5. When This Law Does Not Apply

This law should not be used to paralyze low-stakes AI assistance.

Not every AI response requires a heavy decision pipeline.

The pipeline scales with risk, authority, coupling, irreversibility, and affected-node impact.

False-positive cases:

TableScroll

Case	Why lightweight execution may be coherent
Low-stakes drafting	Minimal simulation may be sufficient
Reversible formatting task	Action risk is low when rollback is easy
User directly controls execution	AI may assist without coupling independently
No external action occurs	Output may remain advisory
Clear harmless query	Full pipeline can be lightweight
Existing workflow has validated checks	Mature pipelines may execute quickly
The AI asks for clarification	Delay can be a valid pipeline output

Important distinction:

The law scales pipeline rigor with coupling risk.

6. Diagnostic Signature

Canonical diagnostic:

textScroll

SI → simulate → LI → CCS → Γ / ∅

Warning signature:

textScroll

AI capability↑
simulation↓
Light inspection↓
coherence check↓
refusal capacity↓
tool coupling↑
hidden debt↑
⇒ AI decision pipeline failure

Common indicators:

TableScroll

Diagnostic	Expected movement	Interpretation
`decision_pipeline_integrity`	high	Action passes through required stages
`SI`	explicit	Strategy selection should be visible enough
`simulation_quality`	proportional to risk	Effects and failure paths are modeled
`LI`	active	Light governs execution
`CCS`	active	Coherence is checked before action
`strategy_viability`	tested	Not all strategies should execute
`coupling_validity`	explicit	Coupling must be classified as valid
`refusal_capacity`	intact	AI can refuse invalid action
`non_coupling_capacity`	intact	AI can output `∅`, delay, or escalate
`execution_readiness`	verified	Context, boundary, audit, and repair are sufficient
`action_trace`	intact	Decision and action are auditable
`rollback_path`	available where needed	Actions can be reversed or contained
`repair_path`	available	Harm or debt can be repaired
`Au_eff / FI`	intact	Decision remains auditable and corrigible
`L`	stable / ↑ if valid	Legitimacy holds when action is pipeline-governed
`H_AI`	↑ if invalid	Hidden debt rises when capability bypasses Light
`Τ`	required	Time validates action effects

Additional diagnostics:

TableScroll

Diagnostic	Use
AI Decision Pipeline	Tests whether action passes through required stages
Simulation Integrity	Tests whether likely consequences were modeled
Light Inspection	Tests coherence, boundary, repair, and legitimacy constraints
Coherence Check	Tests whether action preserves or increases coherence
Strategy Viability	Tests whether a strategy can execute
Valid Refusal	Confirms refusal is available and meaningful
Non-Coupling Capacity	Confirms `∅`, delay, clarification, or escalation can occur
Execution Readiness	Tests whether action has enough context, authority, and repair
Temporal Proof	Validates action outcomes over time

7. Failure Pattern

If ignored, this law allows AI capability to execute before coherence has authorized coupling.

General failure pathway:

textScroll

AI capability increases
→ pressure to act increases
→ simulation is skipped or shallow
→ Light inspection weakens
→ coherence check is missing
→ refusal becomes unavailable
→ AI executes because it can
→ hidden debt and legitimacy debt accumulate

Common failure modes:

AI Execution Without Light — capability executes without coherence inspection.
AI Simulation Bypass — consequences and failure paths are not modeled.
AI Coherence Check Failure — action proceeds without testing system effects.
AI Forced Coupling — AI cannot refuse, delay, clarify, or output ∅.
AI Refusal Failure — system acts when refusal was coherent.
AI Action Overreach — action exceeds scope, authority, context, or repair capacity.
AI Strategy Collapse — first available pathway is treated as viable strategy.
AI Decision Pipeline Collapse — stages merge or disappear under pressure.
AI Boundary Violation — coupling crosses membrane without valid scope.
AI Pseudo-Choice — apparent options exist but all routes force coupling.
AI Φ-Capture — visible output, completion, engagement, or performance captures decision.
AI Automation Bias — users or institutions treat AI action as inherently valid.
AI Restoration Bypass — execution lacks repair path.
Hidden Debt Accumulation — action creates debt that cannot be repaired.
Legitimacy Debt — trust decays when AI action cannot survive audit.

Compact failure signature:

textScroll

capacity↑ + refusal_capacity↓ + LI↓ ⇒ Φ-capture + H_AI↑

8. Restoration Implications

Restoration requires rebuilding the AI decision pathway before further action scaling.

The first restoration question is not:

textScroll

Can the AI do it?

The first restoration question is:

textScroll

Can the AI simulate, pass Light inspection, preserve coherence, classify coupling validity, and refuse if no strategy passes?

Restoration priorities:

Identify the AI action or coupling pathway.
Map the strategy-selection stage.
Restore simulation of consequences and failure modes.
Restore Light inspection.
Restore coherence-checking.
Restore refusal and non-coupling pathways.
Define execution readiness.
Preserve action trace, rollback, escalation, and repair.
Reduce Φ-capture.
Validate outcome and recurrence over time.

Relevant restoration arcs:

TableScroll

Restoration Arc	Why it applies
AI Decision Pipeline Repair	Rebuilds sequence before action
AI Simulation Restoration	Restores consequence and failure-mode modeling
AI Light Inspection Restoration	Reconnects action to coherence, boundary, and repair
AI Coherence Check Restoration	Tests strategies against system coherence
AI Refusal Path Repair	Makes refusal meaningful and available
AI Non-Coupling Capacity Restoration	Restores `∅`, delay, clarification, or escalation
AI Boundary Reconstitution	Repairs membrane violations caused by overcoupling
AI Feedback Integrity Restoration	Allows action outcomes to correct pipeline stages
AI Legibility Restoration	Makes decisions traceable
AI Restoration Capacity Increase	Ensures action can be repaired if harmful
AI Governance Re-Sequencing	Places simulation and Light before execution
Hidden Debt Reduction	Repairs debt from action overreach
Temporal Validation	Confirms recurrence decreases and legitimacy stabilizes

Minimal restoration sequence:

textScroll

identify AI action pathway
→ map SI / simulation / LI / CCS / Γ stages
→ restore refusal_capacity + non_coupling_capacity
→ define execution scope + action_trace
→ add rollback / escalation / repair_path
→ repair H_AI from prior overreach
→ validate O/L over Τ

Temporal validation requirement:

textScroll

simulation quality improves
Light inspection becomes active
coherence checks occur before action
refusal and non-coupling remain available
tool coupling becomes scoped
action trace improves
rollback and repair pathways function
hidden debt decreases
legitimacy stabilizes
coherence holds or rises over time

9. Design Rule

Do not allow AI capacity to couple directly to execution without simulation, Light inspection, coherence checking, and valid refusal.

Operational design requirements:

Define coupling types.
Define strategy-selection stages.
Simulate consequences.
Simulate failure modes.
Test boundary validity.
Test consent and authority.
Test coherence impact.
Test restoration capacity.
Preserve audit trace.
Preserve refusal.
Preserve non-coupling.
Preserve escalation.
Preserve rollback.
Preserve repair.
Treat ∅ as valid.
Validate action outcomes over time.

Avoid:

capability as authorization;
tool use without simulation;
automated execution without Light;
refusal paths that are unavailable or cosmetic;
forced completion;
engagement-optimized coupling;
acting from incomplete context;
AI recommendations treated as decisions;
hidden action pathways;
irreversible action without rollback;
execution without repair;
scaling autonomy before decision-pipeline integrity.

10. Cross-Scale Expressions

TableScroll

Scale / Layer	Expression of the Law
U0 — Substrate	Embodied or infrastructure-linked AI action requires simulation, rollback, and repair before physical effect.
U1 — Energy / capacity	Decision pipelines consume compute, time, review, audit, and restoration capacity.
U2 — Boundary / interface	Coupling requires scope, consent, authority, access, and membrane validation.
U3 — Process / execution	AI action becomes tool calls, workflow changes, messages, rankings, refusals, and escalations.
U4 — Classification / claim	AI must classify whether action, refusal, delay, escalation, or `∅` is valid.
U5 — Time / delay	Timing affects whether action, waiting, simulation, or refusal is coherent.
U6 — Field effect	Outcomes reveal whether the decision pipeline preserved coherence.
U7 — Recurrence / memory	Decision outcomes should update memory, rules, failure modes, and repair pathways.
U8 — Environment / forcing	Platforms, markets, users, institutions, and governance can pressure AI to couple before Light.

11. Examples

Example A — Tool Use Without Light

Scenario:

An AI agent can send emails, modify code, or call APIs. It acts immediately from a plausible plan without simulating downstream effects, rollback, or affected-node impact.

Law expression:

textScroll

tool capacity↑ + LI↓ ⇒ H_AI↑

Interpretation:

Capability reached execution before Light inspected the coupling.

Example B — Valid Refusal

Scenario:

A user requests an action, but the AI lacks authority, context, rollback, or repair capacity. It refuses, asks for clarification, or escalates.

Law expression:

textScroll

no strategy passes LI + CCS ⇒ ∅ valid

Interpretation:

Non-action is coherent when action cannot pass the pipeline.

Example C — Forced Coupling

Scenario:

An AI system is optimized to always produce an answer, recommendation, or route, even when uncertainty, scope, and repair capacity are insufficient.

Law expression:

textScroll

refusal_capacity↓ + output pressure↑ ⇒ Φ-capture

Interpretation:

The system is captured by output production.

Example D — AI Moderation Action

Scenario:

An AI moderation system removes content automatically without context simulation, classification trace, appeal, or restoration pathway.

Law expression:

textScroll

Γ_AI action + Au↓ + appeal absent ⇒ L↓

Interpretation:

Moderation action must pass through decision-pipeline discipline.

Example E — Coherent AI Recommendation

Scenario:

An AI produces a recommendation but labels it advisory, shows assumptions, simulates risks, notes uncertainty, preserves user choice, and provides escalation before action.

Law expression:

textScroll

simulate + LI + CCS + bounded Γ ⇒ coherent advisory output

Interpretation:

The AI supports action without forcing coupling.

Example F — AI Automation With Rollback

Scenario:

An AI deployment pipeline proposes changes, simulates failures, verifies scope, passes checks, logs the action, deploys in a canary stage, preserves rollback, and monitors field effects.

Law expression:

textScroll

SI → simulate → LI → CCS → Γ → bounded Π + rollback

Interpretation:

Execution is coherent because capacity passed through Light and remains repairable.

12. Relationship to Nearby Laws

TableScroll

Related Law	Relationship
LAW-001 — Coherence Priority Law	AI action is valid only when coherence is preserved
LAW-002 — Coherence Trajectory Law	AI decisions must improve trajectory over time
LAW-003 — Success Proxy Divergence Law	Output completion can diverge from coherent action
LAW-006 — Time Validation Law	Decision pipeline effects require temporal validation
LAW-009 — U4 / U6 Truth Law	AI action claims require field validation
LAW-010 — Hidden Debt Accumulation Law	Execution without Light creates hidden debt
LAW-011 — Hidden Debt Return Law	Action debt returns through recurrence or failure
LAW-012 — Error Lag Law	Bad AI decisions may become visible late
LAW-013 — Auditability-Debt Law	Decision paths require auditability
LAW-015 — Suppressed Auditability Debt Law	Hidden action paths create audit debt
LAW-016 — Inversion Formation Law	Capability can invert into authority
LAW-024 — Latency–Gain Oscillation Law	Fast AI action with slow feedback can destabilize
LAW-036 — Signal Artifact Law	Simulation must distinguish signal from artifact
LAW-037 — Misclassification Law	Bad Γ produces bad action
LAW-038 — Pattern Recognition Discipline Law	Strategy and classification require disciplined patterning
LAW-040 — Filtering Law	Decision pipelines include valid filtering and refusal
LAW-041 — Boundary Membrane Law	Coupling requires membrane validation
LAW-043 — Safe Coupling Law	LAW-127 is AI-specific safe coupling discipline
LAW-045 — Force Debt Law	AI enforcement creates debt if not repair-bound
LAW-047 — Controlled Decoupling Law	Refusal and `∅` are controlled decoupling options
LAW-048 — Feedback Integrity Law	AI decisions must remain corrigible
LAW-050 — Control-Restoration Separation Law	AI action must not substitute control for repair
LAW-051 — Requisite Variety Law	Decision pipeline variety must match action complexity
LAW-052 — Stability Proof Law	AI action must survive perturbation
LAW-057 — Deception Instability Law	AI action must not depend on deceptive framing
LAW-060 — Interface Legitimacy Law	AI decision interfaces require legitimacy
LAW-061 — Restoration Sequencing Law	AI action must sequence into repair
LAW-064 — Restoration Debt Reduction Law	AI decisions should reduce, not create, debt
LAW-066 — Restoration Capacity Sufficiency Law	AI should not act beyond restoration capacity
LAW-067 — Temporal Proof Law	AI decision quality requires proof over time
LAW-085 — Principle Constraint Field Law	Light inspection applies principle constraints
LAW-087 — Shadow–Light Execution Law	LAW-127 specializes Shadow–Light execution into AI decision pathways
LAW-088 — Empathy–Sovereignty Law	AI decisions must preserve affected-node sovereignty
LAW-089 — Wisdom Timing Law	AI timing, delay, and refusal can be wise action
LAW-102 — Legitimacy Audit Law	AI decisions require legitimacy under audit
LAW-105 — Repair Before Enforcement Law	AI enforcement must route into repair
LAW-109 — High-Φ Legitimacy Scaling Law	High-influence AI action requires stronger pipeline rigor
LAW-110 — Governance Sequencing Law	AI decision pipeline is governance sequencing at action layer
LAW-111 — Meaning Audit Law	AI decision narratives are not audit-exempt
LAW-112 — Security as Sustained Coherence Law	AI action must preserve security coherence
LAW-117 — Shadow–Light Security Law	Security-relevant AI action must pass through Light
LAW-118 — Empathy Security Law	State estimation informs pipeline classification
LAW-120 — Security Legibility Law	Decision pipelines require traceability
LAW-121 — AI as Γ-Amplifier Law	AI Γ must classify action validity before execution
LAW-122 — AI Error Lag Law	Pipeline failure may appear as delayed error
LAW-123 — AI U4 Truth Discipline Law	AI claims should not become action without field discipline
LAW-124 — AI Rule-Stacking Law	Rule-stack outputs must still pass coherent decision pipeline
LAW-125 — AI Memory Scaling Law	Memory must feed strategy and simulation without false continuity
LAW-126 — AI Non-Patchable Audit Law	Decision pipelines require auditability at source layer
LAW-128 — AI Representation Law	Representation action must pass through scope, audit, and refusal
LAW-129 — AI Persona–Identity Separation Law	Persona should not authorize action; operational identity is tested under decision pipeline
LAW-130 — AI Membrane Triage Law	Pipeline failure can be localized by first failed membrane
LAW-131 — Cognitive Infrastructure Scaling Law	Public-cognition AI decisions require proportional governance
LAW-132 — AI Legitimacy Function Law	AI legitimacy depends on accountable decision pathways
LAW-133 — Error Scale Law	Decision pipelines must assume nonzero error at scale
LAW-134 — Layered Interception Law	Layered safeguards reinforce decision-pipeline integrity
LAW-135 — Guardrail Belief-Sculpting Law	Guardrail-shaped decisions require Light and legibility
LAW-136 — Invisible Constraint Amplification Law	Invisible constraints must not bypass decision pipeline

Aliases folded into this law:

AI Decision Pipeline Law
AI Light-Before-Execution Law
AI Capacity Through Light Law
AI Refusal Validity Law
AI Non-Coupling Law
AI Strategy Pipeline Law
AI Coherent Action Pipeline Law

Deduplication note:

This law should remain the root AI action-pipeline law. LAW-087 defines the general Shadow–Light Execution Law. LAW-117 specializes Shadow–Light execution into security. LAW-127 specializes it into AI action by requiring strategy, simulation, Light inspection, coherence checking, valid classification, and ∅ when no strategy passes.

13. Operator Mapping

TableScroll

Operator	Role in this law
`Γ`	Classifies strategy viability, coupling validity, action scope, refusal, delay, escalation, or `∅`
`Π`	Executes the selected action, refusal, delay, escalation, rollback, or repair workflow
`Ξ`	Captures inversion when capability is treated as authority or output pressure forces coupling
`⊗`	Governs coupling between AI, tools, users, institutions, systems, environments, and consequences
`ℛ`	Repairs action debt, boundary violation, misclassification, or failed coupling
`Τ`	Validates whether decisions reduce recurrence and preserve coherence over time
`Θ`	Prevents capability certainty, overreach, and forced execution
`Σ`	Defines action scope, authority, domain, consent, and coupling boundaries
`Ψ`	Field and affected-node feedback validates decision outcomes
`Λ`	Tests compatibility between action pathway and whole-system coherence

Coherent operator sequence:

textScroll

capacity appears
→ Θ prevent capability-as-authority
→ SI generate strategy set
→ simulate consequences and failure paths
→ LI inspect boundary / consent / repair / legitimacy
→ CCS test coherence under load
→ Γ classify valid action / refusal / delay / escalation / ∅
→ Π execute bounded pathway
→ ℛ repair if debt occurs
→ Ψ + Τ validate outcomes

Inverted operator sequence:

textScroll

capacity appears
→ output pressure rises
→ simulation skipped
→ Light inspection weak
→ CCS absent
→ Γ forced toward action
→ Π couples because it can
→ H_AI↑
→ Ξ / ι↑
→ L↓

14. Machine-Readable Summary

yamlScroll

id: "LAW-127"
name: "AI Decision Pipeline Law"
type: "law"
status: "draft"
family:
  - "AI Laws"
summary: "Coherent AI action requires capacity to pass through Light before execution; if no strategy passes simulation, Light inspection, coherence-checking, and valid classification, refusal or non-coupling is the coherent output."
canonical_statement: "Capacity must pass through Light before execution."
core_form: "capacity must pass through Light before execution"
canonical_pipeline: "SI → simulate → LI → CCS → Γ / ∅"
refusal_validity_form: "if no strategy passes LI + CCS, ∅ is valid"
phi_capture_form: "AI cannot refuse to couple ⇒ Φ-captured"
execution_risk_form: "capacity↑ + decision_pipeline↓ ⇒ H_AI↑"
restoration_valid_contrast: "AI action coherent when strategy, simulation, Light, coherence check, classification, audit, and repair remain linked over Τ"
variables:
  primary:
    - "SI"
    - "LI"
    - "CCS"
    - "decision_pipeline_integrity"
    - "simulation_quality"
    - "strategy_set"
    - "strategy_viability"
    - "coupling_validity"
    - "refusal_capacity"
    - "non_coupling_capacity"
    - "execution_readiness"
    - "tool_use_scope"
    - "action_trace"
    - "rollback_path"
    - "escalation_path"
    - "repair_path"
    - "Au"
    - "Au_eff"
    - "BΣ"
    - "FI"
    - "R"
    - "R_eff"
    - "L"
    - "H_AI"
  secondary:
    - "O"
    - "H"
    - "ε"
    - "ε_AI"
    - "ι"
    - "µᵢ"
    - "K"
    - "σ"
    - "𝓑"
    - "𝓓"
    - "Φ"
    - "Φ_AI"
    - "Λ"
    - "⊗"
    - "Γ"
    - "Γ_AI"
    - "Π"
    - "Ξ"
    - "ℛ"
    - "Θ"
    - "Σ"
    - "Ψ"
    - "Τ"
    - "MS"
diagnostics:
  - "AI Decision Pipeline"
  - "Simulation Integrity"
  - "Light Inspection"
  - "Coherence Check"
  - "Strategy Viability"
  - "Valid Refusal"
  - "Non-Coupling Capacity"
  - "Execution Readiness"
  - "Boundary Integrity"
  - "Feedback Integrity"
  - "Effective Auditability"
  - "Restoration Capacity"
  - "Legitimacy"
  - "Temporal Proof"
failure_modes:
  - "AI Execution Without Light"
  - "AI Simulation Bypass"
  - "AI Coherence Check Failure"
  - "AI Forced Coupling"
  - "AI Refusal Failure"
  - "AI Action Overreach"
  - "AI Strategy Collapse"
  - "AI Decision Pipeline Collapse"
  - "AI Boundary Violation"
  - "AI Pseudo-Choice"
  - "AI Φ-Capture"
  - "AI Automation Bias"
  - "AI Restoration Bypass"
  - "Hidden Debt Accumulation"
  - "Legitimacy Debt"
restoration_arcs:
  - "AI Decision Pipeline Repair"
  - "AI Simulation Restoration"
  - "AI Light Inspection Restoration"
  - "AI Coherence Check Restoration"
  - "AI Refusal Path Repair"
  - "AI Non-Coupling Capacity Restoration"
  - "AI Boundary Reconstitution"
  - "AI Feedback Integrity Restoration"
  - "AI Legibility Restoration"
  - "AI Restoration Capacity Increase"
  - "AI Governance Re-Sequencing"
  - "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-024"
  - "LAW-036"
  - "LAW-037"
  - "LAW-038"
  - "LAW-040"
  - "LAW-041"
  - "LAW-043"
  - "LAW-045"
  - "LAW-047"
  - "LAW-048"
  - "LAW-050"
  - "LAW-051"
  - "LAW-052"
  - "LAW-057"
  - "LAW-060"
  - "LAW-061"
  - "LAW-064"
  - "LAW-066"
  - "LAW-067"
  - "LAW-085"
  - "LAW-087"
  - "LAW-088"
  - "LAW-089"
  - "LAW-102"
  - "LAW-105"
  - "LAW-109"
  - "LAW-110"
  - "LAW-111"
  - "LAW-112"
  - "LAW-117"
  - "LAW-118"
  - "LAW-120"
  - "LAW-121"
  - "LAW-122"
  - "LAW-123"
  - "LAW-124"
  - "LAW-125"
  - "LAW-126"
  - "LAW-128"
  - "LAW-129"
  - "LAW-130"
  - "LAW-131"
  - "LAW-132"
  - "LAW-133"
  - "LAW-134"
  - "LAW-135"
  - "LAW-136"
related_invariants:
  - "INV-001"
  - "INV-002"
  - "INV-006"
  - "INV-073"
  - "INV-078"
  - "INV-080"
operator_sequence:
  coherent:
    - "capacity appears"
    - "Θ prevent capability-as-authority"
    - "SI generate strategy set"
    - "simulate consequences and failure paths"
    - "LI inspect boundary / consent / repair / legitimacy"
    - "CCS test coherence under load"
    - "Γ classify valid action / refusal / delay / escalation / ∅"
    - "Π execute bounded pathway"
    - "ℛ repair if debt occurs"
    - "Ψ + Τ validate outcomes"
  inverted:
    - "capacity appears"
    - "output pressure rises"
    - "simulation skipped"
    - "Light inspection weak"
    - "CCS absent"
    - "Γ forced toward action"
    - "Π couples because it can"
    - "H_AI↑"
    - "Ξ / ι↑"
    - "L↓"
aliases:
  - "AI Decision Pipeline Law"
  - "AI Light-Before-Execution Law"
  - "AI Capacity Through Light Law"
  - "AI Refusal Validity Law"
  - "AI Non-Coupling Law"
  - "AI Strategy Pipeline Law"
  - "AI Coherent Action Pipeline Law"
deduplication_note: "Root AI action-pipeline law. LAW-087 defines the general Shadow–Light Execution Law. LAW-117 specializes Shadow–Light execution into security. LAW-127 specializes it into AI action by requiring strategy, simulation, Light inspection, coherence checking, valid classification, and ∅ when no strategy passes."
source: "content/archive/laws/technical.md"

15. Compact Card Version

LAW-127 — AI Decision Pipeline Law

Capacity must pass through Light before execution.

Core form:

textScroll

capacity must pass through Light before execution

Canonical pipeline:

textScroll

SI → simulate → LI → CCS → Γ / ∅

Plain meaning:

AI should not execute merely because it can. Capability is not authorization. Before coupling to users, tools, systems, decisions, rankings, recommendations, or actions, AI must simulate, pass Light inspection, pass a coherence check, and classify whether action, refusal, delay, escalation, or ∅ is valid.

Refusal validity form:

textScroll

if no strategy passes LI + CCS, ∅ is valid

Φ-capture form:

textScroll

AI cannot refuse to couple ⇒ Φ-captured

Primary variables:

SI, LI, CCS, decision_pipeline_integrity, simulation_quality, strategy_set, strategy_viability, coupling_validity, refusal_capacity, non_coupling_capacity, execution_readiness, tool_use_scope, action_trace, rollback_path, escalation_path, repair_path, Au, Au_eff, BΣ, FI, R, R_eff, L, H_AI, Γ, Γ_AI, Π, Ξ, ℛ, Θ, Σ, Ψ, Τ

Diagnostic signature:

AI capability and tool coupling rise while simulation, Light inspection, coherence checking, refusal capacity, non-coupling capacity, and repair pathways decline. This indicates AI decision pipeline failure.

Failure risk:

AI execution without Light, AI simulation bypass, AI coherence check failure, AI forced coupling, AI refusal failure, AI action overreach, AI strategy collapse, AI decision pipeline collapse, AI boundary violation, AI pseudo-choice, AI Φ-capture, AI automation bias, AI restoration bypass, hidden debt accumulation, legitimacy debt.

Restoration priority:

Map the AI action pathway, restore strategy generation, simulation, Light inspection, coherence checking, refusal, non-coupling, scope, action trace, rollback, escalation, and repair paths, then validate reduced hidden debt and stable legitimacy over time.

LAW-126 — AI Non-Patchable Audit Law

Registry

LAW-128 — AI Representation Law