AI Control Library / CF-12: Security & Abuse Resistance
Designation: Supplemental | Control types represented: Operational and Technical
Control requirement
Revoke AI access promptly when employment, sponsorship, role, task or service need ends. Cover active sessions, tokens, keys, delegated grants and cached permissions across connected systems.
Why this matters
Disabling a primary login may leave provider keys or workload tokens active. Revocation must follow the complete access path to be effective.
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
Apply to departures, role changes, terminated suppliers, completed tasks and suspected compromise. Define risk-based revocation deadlines.
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 identity owner coordinates revocation; HR or sponsors provide triggers; system owners verify downstream removal.
Implementation guidance
- Define lifecycle and emergency revocation triggers.
- Map identities to sessions, credentials and delegated access.
- Revoke grants and invalidate active access within approved deadlines.
- Verify denial and reconcile missed systems or persistent tokens.
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 |
|---|---|
| Departing user | Access is removed within the required interval. |
| Previously issued token | Resource rejects it after effective revocation. |
| Completed agent task | Task credentials cannot be reused. |
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
- Revocation triggers
- Removal logs
- Denial tests
- Exception 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
Disable remaining paths urgently, investigate post-trigger activity and repair the lifecycle integration that missed removal.
Example in practice
Hypothetical example: A contractor’s login is disabled, and a follow-up check discovers a separate model-provider API key. The key is revoked and its activity reviewed.
Related controls
- CF12-S02: AI System Identity Inventory
- CF12-S04: AI System Access Provisioning
- CF14-008: Agentic Task Scope & Graceful Termination
Framework mappings
- CSA AICM (IAM-07)
- NIST AI RMF (MAP 1.1)
- 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 Security & Abuse Resistance
Version 1.0