OpenAI scales tiered cyber AI access, tying power to verified trust

OpenAI is expanding its Trusted Access for Cyber program with new tiers that unlock more capable tools for vetted security teams, including GPT-5.4-Cyber, while emphasizing user verification to prevent misuse. The rollout aims to widely enable defensive cybersecurity work, though access will be gradual and limited to approved vendors and researchers at first, with ongoing talks about government access and ongoing safety reviews.