OpenAI restricts new GPT-5.5-Cyber model to ‘critical cyber defenders’ only

OpenAI is preparing to launch a new cybersecurity-focused artificial intelligence model, GPT-5.5-Cyber, in a highly restricted rollout aimed at strengthening digital defence systems rather than serving the general public.

CEO Sam Altman said on X that the model will be made available “in the next few days,” but only to a select group of trusted “cyber defenders.”

These are expected to include vetted cybersecurity professionals, institutions, and possibly government-linked security agencies.

OpenAI said it will work with “the entire ecosystem and the government” to establish secure and verified access pathways.

Unlike previous public AI releases, GPT-5.5-Cyber will not be widely accessible at launch.

The company has not disclosed technical specifications, but the naming suggests it is a specialised adaptation of the recently released GPT-5.5 model, which OpenAI described as its most advanced and intuitive system to date.

The controlled release reflects a growing trend in the artificial intelligence industry, where frontier models are increasingly being withheld from public access due to concerns over potential misuse.

Cybersecurity, biological research, and advanced reasoning systems are among the most sensitive areas being placed under tighter distribution controls.

OpenAI has previously used staged or restricted deployments for safety-critical models, including systems designed for life sciences and biological research.

The move also comes amid broader industry and government discussions about how to balance innovation with national security risks, particularly as AI becomes more capable of assisting both cyber defence and cyberattacks.

Altman did not specify when or whether GPT-5.5-Cyber would eventually be released to the public.