AI Control Library / CF-04: Auditability & Evidence
Designation: Core | Control types represented: Operational and Technical
Control requirement
Protect AI audit records against unauthorized alteration, deletion and substitution throughout generation, transport, storage and export. Provide integrity verification and preserve an attributable history of authorized corrections and retention actions.
Why this matters
A complete-looking log can be misleading if the same actor who performs an action can erase its evidence. Integrity protection must cover privileged access and exports, not only encryption of the storage device.
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
Logs supporting security investigations, consequential decision review and assurance, including collectors, archives and evidence copies.
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 logging platform owner implements safeguards; security defines trust boundaries and evidence custodians verify protected exports.
Implementation guidance
- Identify threats to record integrity across sources, collectors, storage, administrative roles and backups; separate modification authority where feasible.
- Protect transport and storage, using access controls, independent collection and appropriately configured retention or tamper-evident mechanisms.
- Record deletion, retention changes and corrections through a protected audit path; preserve originals and distinguish a corrected view from the source evidence.
- Verify integrity and completeness assumptions during retrieval and export, testing privileged deletion and collection-loss scenarios.
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 |
|---|---|
| Attempt unauthorized log modification in a controlled test | The change is prevented or detected through a protected independent mechanism. |
| Inspect an exported evidence set | Its origin, integrity reference and handling history can be verified. |
| Exercise authorized retention deletion | The action follows policy and leaves the required accountability record. |
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
- Log protection and trust-boundary design
- Retention and administrative configuration
- Integrity and deletion tests
- Protected export and correction records
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 suspected tampering to security and the investigation authority. Preserve independent copies and limit reliance on affected records until their integrity and gaps are assessed.
Example in practice
Hypothetical example: An application administrator can clear local logs. Events are also sent to a separately controlled repository, and a test confirms that deleting the local copy does not remove the evidence of the action.
Related controls
- CF04-011: AI Audit Log Access Control & Accountability
- CF04-013: AI Monitoring System Failure Reporting
- CF04-016: AI Governance Evidence Management
Framework mappings
- CSA AICM (LOG-02, LOG-09)
- NIST AI RMF (MANAGE 4.1)
- EU AI Act (Art. 12(1))
- ISO 42001 (Cl. 9.1)
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