Decision Signals | 10 November 2026 | The Audit Trail Is Not the Decision Trail
Decision Signals · Issue 04The Decision Layer™ · Week 2

The Audit Trail Is Not the Decision Trail.

Five signals that an organisation may reconstruct AI activity without being able to explain the risk, mitigation and decision behind it.
Published10 November 2026
Read time8 minutes
CycleWeek 2 of 4
DI focusTraceability
01 · The distinction

Logs tell you what happened. A decision trail has to explain why.

An organisation can have excellent technical logging and weak Decision Traceability. The difference shows up when someone asks a question the log was never designed to answer.

Audit trail

Reconstructs activity.

Who logged in. Which model responded. What transaction occurred. Which workflow fired. Which user clicked approve.

Decision trail

Reconstructs judgement.

What evidence existed. Which assumptions mattered. What AI contributed. What the human decided. What was challenged. Why the choice made sense.

02 · Research signal

Disclosure, mitigation and assurance answer different questions.

Research from the MIT AI Risk Initiative and MIT FutureTech examines how major companies publicly acknowledge AI risks and describe their responses. One reported concern is the limited connection between disclosed risks and the mitigations intended to address them.

Important boundary: Public disclosure is not a reliable proxy for internal governance effectiveness. An undisclosed control may operate effectively; a disclosed mitigation may not. The board's task is to test the evidence behind consequential choices.
01 · Visibility

Risk disclosure

What exposure does management acknowledge? This establishes visibility, not control effectiveness.

Board challenge: What material risks remain uncertain?

02 · Operation

Mitigation evidence

Which measures address each exposure, and what testing shows they work under relevant conditions?

Board challenge: Can the risk-to-control link be demonstrated?

03 · Accountability

Decision assurance

Who chose the response, on which evidence, with what authority, challenge and residual-risk acceptance?

Board challenge: Could an independent reviewer reconstruct the judgement?

Risk→Consequence→Mitigation→Decision→Authority→Evidence→Verification
A useful assurance chain. Each link requires evidence; none is established merely because the next one is documented.

Research: MIT AI Risk Initiative and MIT FutureTech, Unmitigated? How Are the World's Largest Companies Responding to AI Risks? See MIT AI Risk Disclosure Tracker. The decision-assurance interpretation above is The Decision Layer™ analysis, not a conclusion attributed to MIT.

03 · Five decision signals
S1 · The model version changed

The output can be preserved while the decision context disappears.

What changed?

The provider, model version, or configuration changed after the original decision.

Decision consequence

Replaying the workflow today may produce a different answer, so current evidence cannot safely substitute for historical evidence.

Reopening question

Do we preserve the model and configuration context needed to reconstruct the original decision?

S2 · The prompt survived. The policy threshold did not.

Technical context is not organisational context.

What changed?

Prompts and outputs are logged, but the risk appetite, approval threshold, or business rule in force at the time is not linked to them.

Decision consequence

The organisation can reproduce the interaction without explaining why the output was acceptable.

Reopening question

Can we connect the technical record to the authority and policy that governed the decision?

S3 · The human override is logged. The challenge is not.

The action survives. The reasoning does not.

What changed?

A reviewer overrode or accepted an AI recommendation, but only the final action was retained.

Decision consequence

The organisation cannot distinguish meaningful human judgement from ceremonial approval.

Reopening question

What evidence shows what the human actually considered?

S4 · The supplier boundary breaks the trail

The decision crosses systems. The evidence does not.

What changed?

An external AI service supplies a score, recommendation or agent action, but the underlying evidence remains with the supplier.

Decision consequence

Internal logs begin after the most important reasoning has already occurred.

Reopening question

What traceability rights are contractual, and what evidence disappears at the API boundary?

S5 · The decision moved across three tools

Each system has a log. Nobody has the story.

What changed?

Recommendation, approval, and execution now happen in different systems owned by different teams.

Decision consequence

Local logs remain intact while end-to-end accountability becomes fragmented.

Reopening question

Can one decision identifier connect evidence, authority, action and outcome?

04 · The reconstruction test

Pick one decision from six months ago.

  1. Who initiated it?
  2. What evidence was available?
  3. Which model, configuration and policy context applied?
  4. What did the AI contribute?
  5. What did the human contribute?
  6. What challenge occurred?
  7. Who had authority?
  8. What would have caused the decision to be reopened?
  9. Which material AI risk and business consequence prompted the decision?
  10. Which mitigation was selected, and what evidence showed it addressed that risk?
  11. Who accepted the residual risk, and where is that acceptance recorded?

If the answer requires days of archaeology across email, meeting notes, logs and people's memories, the organisation may have logging. It may not have Decision Traceability.

Executive test · next 30 days

Select one consequential AI-enabled decision from the last six months. Ask management to produce the risk-to-mitigation mapping, the contemporaneous decision record, and outcome verification. Give an independent reviewer the evidence pack without the original decision-maker in the room. Record missing links, accountable owners, and deadlines for remediation.

Decision Infrastructure View
Component affected
TRACEABILITY
Decision signal
The organisation can reconstruct activity but not authority, evidence, reasoning and challenge end to end.
Decision consequence
Assurance and accountability weaken precisely when a consequential outcome is challenged.
Evidence needed
A connected decision trail linking participants, evidence, authority, AI contribution, human judgement, challenge, action and outcome.
Reopening question
Has the system, model, policy or decision path changed enough that the old reconstruction is no longer reliable?
Research implication
Public disclosure may not show how identified risks relate to mitigations; internal decision assurance must be assessed separately.
Executive action
Reconstruct one consequential AI decision, including its risk, mitigation, authority, residual-risk acceptance and verification, within 30 days.
The audit trail is not the decision trail.

A risk disclosure shows what management acknowledges. Mitigation evidence shows what was done. A decision trail shows why the choice was defensible at the time.

Next: Board Case Study, 17 November. The Evidence Did Not Survive.