Rolling record / ARQ 78 Severe · ▲ +24 7-day 314 events in window / as of 14 Sept 2026
ARQAI Risk Quotient
capability 70 Severe

OpenAI releases GPT-5.5 alongside its strongest safeguards to date

A major capability step and a deliberate safety one — fuller red-teaming and stricter cyber classifiers shipped with the model.

A major capability step and a deliberate safety one — fuller red-teaming and stricter cyber classifiers shipped with the model.

Risk 3 · Mitigation 1.