Comment by International AI Safety Report 2026

Frontier AI Safety Frameworks commonly require an initial capabilities evaluation prior to risk mitigation, as well as a residual risk analysis or a safety case, often informed by red-teaming, after mitigation.