AI Control Library / CF-04: Auditability & Evidence
Designation: Core | Control types represented: Operational and Technical
Control requirement
Monitor AI audit events for meaningful anomalies and route them to assigned responders with defined triage, investigation and containment actions. Validate detection coverage and preserve the event context needed to assess an alert.
Why this matters
A logging platform does not provide protection if significant events remain unread or alerts have no owner. Detection should connect an observable pattern to a realistic response rather than generate volume without decisions.
How the control types work together
Operational controls define scope, authority and the actions required when conditions are not met. Technical controls produce evidence or enforce the relevant boundaries. Both must apply to the same deployment and operating conditions.
When it applies
AI security and accountability events, including authorization failures, unusual tool activity, privilege changes, log tampering and sensitive-data access.
Define the assessed boundary, dependencies and evidence limitations. Tailor depth and review frequency to potential impact, and document why a particular safeguard or exclusion is appropriate to the actual deployment.
Ownership
The security monitoring owner maintains detection and response; AI system specialists provide workflow context and authority boundaries.
Implementation guidance
- Derive detection cases from threat scenarios and prohibited behavior, distinguishing ordinary variation from events needing investigation.
- Define event sources, correlation, thresholds or rules, recipients, coverage hours and response expectations based on the potential harm.
- Provide responders with appropriate context and approved containment routes while restricting sensitive log access.
- Exercise detections and handoffs, review false positives and missed scenarios, and update rules after material workflow or threat changes.
How to verify it
Use representative records and authorized controlled tests. Identify the relevant version, operating conditions and expected result before testing, and retain enough evidence to support the conclusion.
| Test | Expected result |
|---|---|
| Generate a harmless representative anomaly | The intended detection produces an actionable alert for the assigned team. |
| Trace an alert through triage | Investigation and containment decisions are recorded with the relevant evidence. |
| Inspect a change in agent permissions | Detection coverage and expected behavior are reconsidered. |
Record a finding when required results cannot be demonstrated or practice differs materially from the approved conditions. Missing evidence limits the conclusion; it should not be treated as a successful test.
Evidence to retain
- Detection cases and rule configuration
- Alert routing and response procedures
- Exercises and investigation records
- Tuning and coverage reviews
Protect evidence according to its sensitivity, access needs and retention rules. Avoid unnecessary copies of personal information, secrets or confidential material, while preserving the records needed for authorized review.
If the control fails
Escalate credible material misuse to incident response and the system owner. Contain the affected identity or action path under approved authority and address unowned or ineffective detections.
Example in practice
Hypothetical example: An agent begins issuing repeated denied requests to a new tool. Correlated authorization and execution events prompt investigation, revealing an unsafe configuration change before additional access is granted.
Related controls
- CF04-008: AI System Logging Scope & Event Definition
- CF04-013: AI Monitoring System Failure Reporting
- CF09-005: AI Security Monitoring & Alerting
Framework mappings
- CSA AICM (LOG-05)
- NIST AI RMF (MANAGE 4.1)
- EU AI Act (Art. 12(1))
- ISO 42001 (Cl. 9.1)
- MITRE ATLAS
Review and maintenance
Review at the interval approved by the control owner and after relevant changes, incidents or failed assessments. Update affected procedures and configuration, address evidence gaps and verify corrective action before closing findings.
Back to Auditability & Evidence
Version 1.0