Supplements (4.x–12.x)
The AI-OGF supplement system provides implementation‑ready guidance for each major control family.
Supplements expand the core requirements with:
- Detailed implementation steps
- Domain‑specific considerations
- Examples and applied scenarios
- Extended controls and operational patterns
- Notes, exceptions, and environment‑specific variations
All supplements are located in:
docs/supplements/
Each supplement folder (4.x–12.x) contains the documents for that control family.
Linked Supplements
4.x — Principles
- 4.1 Transparency
- 4.2 Least Autonomy
- 4.3 Human Override
- 4.4 Isolation by Design
- 4.5 Dependency Awareness
- 4.6 Fail‑Safe Defaults
- 4.7 Workflow Layer Limits
- 4.8 Constrained Autonomy Envelope (CAE)
- 4.9 Compensating Controls for Life‑Determinant AI
- 4.10 Identity‑Bound Authority
5.x — Dependency Mapping
- 5.2.1 Map AI → Infrastructure
- 5.2.2 Map AI → AI
- 5.2.3 Map AI → Humans
- 5.2.4 Map Humans → AI
- 5.2.5 Circular Dependencies
- 5.3 Placeholder
- 5.4 Placeholder
6.x — Autonomy & Identity Controls
- 6.2.1 Define Autonomy Levels
- 6.2.2 Autonomy Boundary Enforcement
- 6.2.3 Autonomy Drift Detection
- 6.2.4 Autonomy Escalation Rules
- 6.3 Decision Authority & Escalation
7.x — Continuity and Safety Controls
- 7.2.1 Continuity Boundaries
- 7.2.2 Fallback Behavior
- 7.2.3 Drift Detection
- 7.2.4 Monitoring and Telemetry
- 7.2.5 Escalation Paths
- 7.2.6 Safe Mode Behavior
- 7.2.7 Destructive Action Controls
8.x — AI Workflow Layer Limits (AI‑WLL)
- 8.1 Purpose
- 8.1.1 Rationale (contained within 8.1)
- 8.1.2 Applicability (contained within 8.1)
- 8.2.1 Maximum Workflow Depth
- 8.2.2 Mandatory Human Checkpoints
- 8.2.3 Prohibited Recursive AI Calls
- 8.3.1 Cross‑AI Interaction Limits
- 8.3.2 Workflow Telemetry Requirements
- 8.3.3 Override & Rollback Requirements
- 8.4.1 Infrastructure Deployment Example
- 8.4.2 Remediation Example
- 8.5.1 Exceptions
- 8.5.2 Environment‑Specific Variations