OpenAI announced it is rolling out GPT-5.5-Cyber — a fine-tuned version of GPT-5.5 optimized for defensive cybersecurity — to vetted critical infrastructure operators through its Trusted Access for Cyber (TAC) program. The UK AI Security Institute's independent evaluation, published the same day, found GPT-5.5 achieves a 71.4% pass rate on expert-level cyber challenges — above Claude Mythos Preview (68.6%) and GPT-5.4 (52.4%) — and completed the "Last Ones" 32-step corporate network attack simulation end-to-end in 2 of 10 attempts, making it only the second model to do so. Access remains restricted to vetted defenders via application at chatgpt.com/cyber, with OpenAI treating GPT-5.5's cybersecurity capabilities as "High" under its Preparedness Framework, below the Critical threshold for autonomous zero-day exploit development.