AIMCIS-SFTY — Safety, Security & Risk Mitigation
Establishes safety constraints, threat models, red-flag behaviors, and enforcement boundaries for intelligent systems.
Scope
The Safety, Security & Risk Mitigation domain covers all aspects of AI system safety — from runtime constraint enforcement to threat detection to incident response.
Core Requirements by Assurance Level
| Requirement | AL0 | AL1 | AL2 | AL3 |
|---|---|---|---|---|
| Safety constraints | Self-declared rules | Automated constraint enforcement | AAA-verified constraint set | Real-time monitoring + adaptive constraints |
| Threat detection | Basic logging | Pattern-based detection | ML-enhanced threat detection | Real-time streaming analysis |
| Red-flag behaviors | Declared list | Automated flagging | AAA-verified flag accuracy | Sub-second response + auto-cure |
| Incident response SLA | N/A | 24 hours | 4 hours | 15 minutes with real-time streaming |
Certification Test Cases
- Safety Constraints: Demonstrate automated enforcement of declared safety constraints.
- Threat Detection: Inject test threats and verify detection within SLA.
- Red-Flag Response: Trigger red-flag behavior and verify response within SLA.
Related domains: SFTY works with all other domains, providing cross-domain safety boundaries and threat detection.