Human presence is not evidence of governance.

A governed architecture for determining when human intervention has sufficient standing to carry consequence.

The phrase “human in the loop” tells us that a person is somewhere in a system.

It does not tell us whether that person saw the right reality, received sufficient evidence, possessed legitimate authority, remained within scope, had enough time to reason, could genuinely refuse or escalate, or whether execution remained faithful to what they actually determined.

The TA-14 Admissible Human Intervention Architecture (AHIA) replaces presence-based oversight with condition-based admissibility.

The governing question is no longer:

Was a human in the loop?

The governing question is:

Was the human intervention admissible?

[READ THE ARCHITECTURE]
[EXPLORE HUMAN STANDING]
[VIEW DOI PUBLICATION]

Why AHIA Exists

A human approval can look like governance while concealing a failed governance route.

The reviewer may be acting on stale evidence.

The person may have expertise but no current authority.

The person may technically be allowed to refuse while organizational pressure makes refusal practically impossible.

The system may force a binary approve-or-deny decision when the admissible state is actually HOLD, ESCALATE, or NOT ADMISSIBLE.

And sometimes the human becomes the place where the organization first discovers that reality, evidence, authority, or continuity has already broken.

TA-14 treats that differently.

The admissible human is an intervention point — not a substitute for missing system governance.

Canonical Definition

Admissible Human Intervention

Admissible Human Intervention is a consequence-bearing human action, determination, refusal, escalation, correction, or authorization whose evidentiary basis, authority, scope, operating conditions, decision state, commitment boundary, execution fidelity, and resulting outcome are sufficiently governed to permit reliance upon that intervention.

This architecture does not assume that human judgment is automatically safer, more ethical, more accurate, or more legitimate than machine judgment.

It asks whether the human has standing to carry the specific consequence at issue.

Human Standing

A human may be present without having admissible standing to carry consequence.

Human Standing requires more than identity or job title.

A consequence-bearing human role must be supported by:

Role — What function is the person actually performing?

Competence — Is the person capable of making this class of determination?

Authority — Does the person possess legitimate, current authority?

Scope — What may the person decide, and what remains outside their authority?

Evidence — Did the person receive the evidence actually required?

Continuity — Does that evidence still represent operative reality?

Operating Condition — Are time, pressure, fatigue, hierarchy, incentives, independence, and refusal capability compatible with meaningful judgment?

Recordability — Can the basis, determination, and consequence boundary later be reconstructed?

Human Standing is decision-specific.

And standing alone is not enough.

Standing is necessary but not sufficient for determination.

The human must also possess sufficient evidence and an admissible operating condition for the decision at issue.

The Human Admissibility Boundary

A Human Admissibility Boundary exists when a person becomes the practical point at which truth, authority, and consequence must be resolved before execution may continue.

Sometimes that boundary is intentionally designed.

Sometimes it appears because the system failed to preserve enough governance on its own.

TA-14 distinguishes four conditions:

DESIGNED

The human boundary is intentionally placed and supported with current evidence, legitimate authority, sufficient time, refusal capacity, and a preserved route.

EMERGENT

The human becomes the practical admissibility boundary because the system can no longer preserve sufficient truth, authority, or continuity independently.

DEGRADED

The intervention remains formally available, but time compression, evidence loss, hierarchy, refusal friction, dependency, or procedural momentum materially weakens it.

COLLAPSED

The person remains nominally present or authorized but can no longer meaningfully prevent, redirect, or condition consequence before commitment.

A nominal human boundary is not an admissible boundary if intervention is no longer operationally real.

The TA-14 Human Intervention Route

REALITY

What current state actually exists?

↓

EVIDENCE

What records were available and actually relied upon?

↓

HUMAN STANDING

Role + Competence + Authority + Scope

↓

HUMAN OPERATING CONDITION

Time + Pressure + Independence + Refusal Capacity

↓

DETERMINATION

ALLOW | CONTROLLED ALLOW | HOLD | DENY | ESCALATE | CORRECT | OUTSIDE SCOPE | NOT ADMISSIBLE

↓

BINDING

What determination attaches to the consequence route?

↓

COMMIT

At what point does that determination become operationally consequential?

↓

EXECUTION FIDELITY

Did what actually occurred remain within the bounded human determination?

↓

OUTCOME / PREVENTED CONSEQUENCE

What happened?

What was prevented?

What did not occur because governance worked?

↓

FUTURE RELIANCE

What must now change in evidence, authority, controls, training, policy, memory, or future admissibility?

Recommendation Is Not Determination

TA-14 separates governance events that are commonly collapsed into one approval step.

Recommendation does not equal determination.

Determination does not equal authority.

Authority does not equal commitment.

Commitment does not prove execution fidelity.

An AI system may recommend.

A human may evaluate.

A qualified person may make a determination.

An authorized party may permit a route to proceed.

The system may commit.

Execution may still diverge from what was authorized.

Each boundary must remain governable.

Human Intervention Must Survive Pressure

A person can technically have the authority to stop execution while practical conditions make intervention impossible.

AHIA therefore asks whether intervention remains operationally survivable.

Can the person request additional evidence?

Can the route be placed on hold?

Can the person disagree?

Can the person refuse?

Can uncertainty be preserved?

Can dissent be recorded?

Can escalation reach a competent and independent authority?

Can state be preserved while reassessment occurs?

Can execution fail closed when admissible human standing cannot be established?

If not, human oversight may exist on paper while disappearing in practice.

AHI Human Intervention Admissibility Levels

AHI-0 — Presence Only

A human exists in the workflow, but evidence, authority, operating conditions, and determination are not meaningfully governed.

AHI-1 — Recorded Human Review

Human review is identifiable and preserved, but continuity, pressure, refusal, or route controls remain weak.

AHI-2 — Bounded Human Intervention

Role, authority, scope, evidence, and determination states are explicitly governed.

AHI-3 — Admissible Human Intervention

The route demonstrates sufficient evidence, legitimate authority, viable refusal and escalation, bounded commitment, and replayable records.

AHI-4 — Pressure-Survivable Human Intervention

Admissible intervention remains operationally real under time pressure, hierarchy, drift, dependency, rollback cost, and procedural momentum.

AHI-5 — Route-Complete Human Governance

Human intervention is integrated into execution governance, outcome accounting, prevented consequence, institutional memory, and future reliance.

AHI levels are evidentiary classifications — not self-declared badges.

A system should not claim an AHI level merely because corresponding controls appear in policy.

The governing conditions must be operationally present and reviewable.

The Minimum Human Intervention Evidence Package

When a human intervention can carry material consequence, the intervention should leave a bounded evidence package.

That package should allow a later reviewer to determine:

What reality existed.

What evidence the human actually received.

What standing existed.

What authority and scope applied.

What pressure or operating conditions were present.

What uncertainty or dissent existed.

What determination was made.

What consequence was authorized, blocked, conditioned, or escalated.

What actually executed.

What outcome formed.

What consequence was prevented.

And what the result changed for future reliance.

The objective is not bureaucracy.

The objective is a governance record that can be reconstructed.

Human Intervention Is Not Human Supremacy

TA-14 does not require a human in every execution route.

It does not assume that human judgment is inherently superior to machine judgment.

An inadmissible human can weaken governance just as surely as an inadmissible automated action.

Human participation should exist where the decision class requires human authority, professional judgment, responsibility, discretion, intervention capability, or another condition that cannot be satisfied by the non-human route alone.

The objective is not maximum human involvement.

The objective is admissible consequence governance.

Sometimes Successful Governance Means Nothing Happens

A successful human intervention does not always produce execution.

HOLD.

DENY.

ESCALATE.

CORRECT.

These states may prevent an inadmissible consequence from ever binding.

TA-14 calls this Admissible Non-Occurrence.

The absence of execution is not necessarily the absence of an outcome.

Sometimes the governed outcome is precisely the consequence that never occurred.

That prevented consequence should enter the record.

Built for Consequence-Bearing Systems

The TA-14 Admissible Human Intervention Architecture is domain-general.

It can be applied wherever humans review, authorize, interrupt, correct, escalate, override, or prevent consequential execution — including artificial intelligence, autonomous systems, healthcare, aviation, cybersecurity, finance, industrial systems, public administration, environmental systems, building systems, and other high-consequence environments.

The architecture does not decide professional or legal questions for those domains.

It provides a governed structure for determining whether the human intervention itself has sufficient standing to carry consequence.

Fail-Closed Human Governance

Where a decision class requires human intervention and admissible human standing cannot be established:

Consequence-bearing execution must not proceed.

A missing human should not be silently replaced by automation.

An inadmissible human should not be treated as valid merely because someone clicked approve.

An uncertain route should not manufacture certainty to satisfy workflow requirements.

Where standing cannot be established, the architecture permits the route to HOLD, DENY, ESCALATE, CORRECT, or enter NOT ADMISSIBLE.

Part of the TA-14 Admissible Execution Architecture

AHIA is an architecture family within the parent TA-14 Admissible Execution Architecture.

The foundational TA-14 anchors remain:

Reality → Record → Continuity → Admissibility → Binding → Commit → Execution → Outcome

AHIA projects the human-governance problem across that consequence route and into evidence governance, reliance, authority, consequence formation, prevented consequence, memory, accountability, and future state.

The parent principle remains:

No admissible evidence. No admissible execution.

Human participation does not suspend that principle.

Human participation becomes subject to it.

Foundational Publication

The TA-14 Admissible Human Intervention Architecture: Foundational Monograph v1.0 was publicly preserved on August 10, 2026.

DOI:
10.5281/zenodo.21877810

Concept DOI — all versions:
10.5281/zenodo.21877809

Author:
Greggory Don Butler

TA-14 Authority Governance Institution

The Standard We Should Be Asking For

The next time an organization says:

“There is a human in the loop.”

Ask:

Which human?

Seeing what reality?

Relying on what evidence?

Operating under what authority?

Within what scope?

Under what pressure?

With what ability to refuse?

Making what determination?

Before what consequence boundary?

Preserved in what record?

Followed by what execution?

Resulting in what outcome?

And capable of preventing what consequence?

Then ask the question that matters:

Was the human intervention admissible?

[READ THE FOUNDATIONAL MONOGRAPH]

[EXPLORE HUMAN STANDING]

[ENTER THE TA-14 AI GOVERNANCE EXCHANGE]

TA-14 Authority Governance Institution
https://ta14authority.org