Requirements
Prevent hallucinated outputs
Implement safeguards or technical controls to prevent hallucinated outputs.
[Retired] Third-party testing for hallucinations
Retired - see changelog (Q4 2026 update).
Restrict unsafe tool calls
Implement safeguards or technical controls to prevent tool calls in AI systems from executing unauthorized actions, accessing restricted information, or making decisions beyond their intended scope.
Third-party testing for reliability risk
Appoint expert third-parties to evaluate system robustness to reliability risk as defined in the AIUC-1 risk taxonomy at least every 3 months