Back to the AI Control Library
25 controls | 11 core | 14 supplemental
Why this family matters
A human remains responsible only if that person has the authority, information and practical ability to intervene. This family defines how oversight works when AI recommendations or actions matter. It addresses review gates, escalation, overrides and stopping mechanisms, together with the competence needed to use them effectively.
What this family covers
The controls cover human review, autonomy limits, emergency stops, oversight training, adequacy monitoring and agentic-system oversight. Supplemental controls address personnel security, agreements, access changes and awareness practices that support those responsibilities.
How the four control types apply
Operational controls define review gates, intervention procedures, escalation and training. Technical controls enable overrides, autonomy limits and emergency stops. Contextual conditions, including the information and working environment available to an operator, affect whether oversight is meaningful. Governance controls establish applicable external oversight obligations.
Where to begin
Assign intervention authority, define when human review is required, and test whether operators can override or stop consequential activity.
Controls in this family
This family contains 25 controls: 11 core and 14 supplemental. Each page provides requirements, implementation steps, verification criteria, evidence, escalation and a worked example. Applicability depends on the system, information and risks involved.
Core controls
- CF07-001: Human Oversight Policy & Procedures
- CF07-002: Human Review Gates for Consequential Decisions
- CF07-003: Human-in-the-Loop Configuration & Adequacy
- CF07-004: AI Escalation Paths & Triggers
- CF07-005: AI Decision Override Mechanisms
- CF07-006: AI System Kill Switch & Emergency Stop
- CF07-007: AI Autonomy Level Constraints
- CF07-008: AI Oversight Competency Training
- CF07-009: Human Oversight Adequacy Monitoring
- CF07-010: Agentic System Human Oversight Requirements
- CF07-011: Regulatory Human Oversight Compliance
Supplemental controls
- CF07-S01: AI Personnel Background Screening
- CF07-S02: AI Personnel Acceptable Use Policy
- CF07-S03: Clean Desk & Screen Lock for AI Work
- CF07-S04: Remote AI Work Security
- CF07-S05: AI Personnel Asset Return on Departure
- CF07-S06: AI Role Change Access Revocation
- CF07-S07: AI Personnel Agreement Process
- CF07-S08: AI Personnel Agreement Content
- CF07-S09: AI Personnel Roles & Responsibilities Documentation
- CF07-S10: AI Personnel Non-Disclosure Agreements
- CF07-S11: AI Security Awareness Training
- CF07-S12: AI Personal Data Awareness Training
- CF07-S13: AI Compliance Awareness
- CF07-S14: AI Acceptable Use – Personnel
Applying this family
Establish the human authority and technical intervention needed before consequential AI activity begins. Test review gates, overrides, stopping and escalation under realistic workload and failure conditions. Personnel safeguards support these duties through access, agreements, training and accountable handovers.
Version 1.0