AI Control Library / CF-04: Auditability & Evidence
Designation: Core | Control types represented: Operational
Control requirement
Establish an AI logging and monitoring policy defining objectives, required event coverage, privacy and retention rules, access, integrity, review responsibilities and response to logging failure. Connect logging requirements to evidence and operational decisions.
Why this matters
More logs do not automatically mean better accountability. Unfocused collection can expose sensitive information while missing the tool action, approval or permission change needed to investigate an incident.
How the control types work together
Operational controls establish the assigned responsibilities, review processes and response decisions described here. Their effectiveness depends on performing the work and using evidence to guide action, rather than approving a document alone.
When it applies
AI applications, agents, models and supporting platforms that generate evidence for security, oversight, assurance or recovery.
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 or observability policy owner maintains requirements with privacy and system owners; operators implement collection and review.
Implementation guidance
- Define investigation and oversight objectives and the event categories needed to support them, including identity, authorization, actions and configuration changes.
- Set rules for content minimization, sensitive-data handling, retention, access and permitted use of collected records.
- Assign collection, monitoring, investigation and evidence-preservation duties, including ownership of supplier visibility gaps.
- Require coverage testing and failure reporting before consequential release and after relevant system or logging 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 |
|---|---|
| Trace a material investigation scenario | The policy requires the events needed to examine it. |
| Inspect prompt or output collection | Its scope and retention have a documented purpose and privacy treatment. |
| Simulate loss of required logging | An owner receives the failure and applies the defined response. |
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
- Approved logging and monitoring policy
- Event and privacy requirements
- Role and retention decisions
- Coverage and failure-response tests
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 unsupported evidence requirements, excessive sensitive collection or unreported logging loss to system, security and privacy owners. Restrict affected operation or collection until acceptable conditions are restored.
Example in practice
Hypothetical example: A team stores every full prompt but omits tool authorization decisions. Policy review reduces content exposure and adds the action and permission events needed to determine whether the agent exceeded its authority.
Related controls
- CF04-008: AI System Logging Scope & Event Definition
- CF04-009: AI Input & Output Logging
- CF04-013: AI Monitoring System Failure Reporting
Framework mappings
- CSA AICM (LOG-01)
- NIST AI RMF (GOVERN 1.1)
- EU AI Act (Art. 12)
- ISO 42001 (Cl. 8.2)
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