OpenAI treats GPT-5.5 biology and cyber capabilities as "High" risk

In the GPT-5.5 system card (April 23, 2026; updated April 24 for API safeguards), OpenAI classifies the model's biological/chemical and cybersecurity capabilities as "High" under its Preparedness Framework and ships what it calls its strongest set of safeguards to date. OpenAI says it ran full predeployment evaluations and targeted red-teaming with roughly 200 early-access partners; GPT-5.5 did not reach the "Critical" cybersecurity level but is a step up from GPT-5.4.