AI Control Library / CF-07: Human Oversight & Escalation
Designation: Supplemental | Control types represented: Operational and Technical
Control requirement
Protect visible and unattended AI work through screen locking, appropriate handling of printed or displayed content and secure workspace practices. Address local devices, shared spaces and remote screen-sharing paths.
Why this matters
Protected data can be exposed through an unlocked terminal or shared display even when application permissions are correct. Generated output may also reveal sensitive information that staff did not expect to appear.
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 workstations, consoles, meeting displays, printed material and remote collaboration where unauthorized people could observe information or act through an open session.
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
Workplace and endpoint security define controls; managers and personnel maintain practices and report exposure.
Implementation guidance
- Define handling expectations for screens, printouts and unattended sessions according to data sensitivity and workspace conditions.
- Configure suitable automatic lock and authentication, with usable manual locking and secure disposal of temporary material.
- Limit unnecessary screen or clipboard sharing and protect displays during meetings or work in shared environments.
- Test configurations and representative practices, reviewing exceptions such as shared operational consoles through explicit compensating controls.
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 |
|---|---|
| Leave a test session unattended | Approved locking and reauthentication operate. |
| Review a screen-sharing scenario | Unnecessary sensitive windows and notifications are not exposed. |
| Inspect temporary printed material | Storage and disposal follow the handling requirement. |
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
- Workspace and endpoint standard
- Lock configuration and tests
- Handling and training records
- Exception and exposure 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
Report suspected exposure or unauthorized session use to security. Secure the session and material, assess what was visible or accessible and correct the affected practice.
Example in practice
Hypothetical example: An operator shares a model dashboard during a meeting. A protected view hides full prompts and the endpoint locks when unattended, reducing accidental exposure without relying on a reminder alone.
Related controls
- CF05-012: AI Sensitive Data Protection
- CF07-S04: Remote AI Work Security
- CF07-S11: AI Security Awareness Training
Framework mappings
- CSA AICM (HRS-03)
- NIST AI RMF (GOVERN 5.2)
- ISO 42001 (Cl. 7.2)
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 Human Oversight & Escalation
Version 1.0