OpenAI Restricts GPT-5.5-Cyber to Cybersecurity Experts

OpenAI is set to launch an advanced cybersecurity model, GPT-5.5-Cyber, intended exclusively for a select group of "cyber defenders." This initiative, announced by CEO Sam Altman, will not be made available to the general public but rather deployed in a targeted manner to help institutions strengthen their defense systems.

The rollout of GPT-5.5-Cyber will begin "in the coming days," according to Altman, who shared this information on the social network X. OpenAI will collaborate with various stakeholders in the ecosystem and the government to define the access criteria for this specialized model.

While details about the initial beneficiaries of this model remain unclear, previous launches of OpenAI models have generally involved verified professionals and institutions. No technical information or detailed specifications have yet been published regarding GPT-5.5-Cyber, although its name suggests that it is a specialized version of the recently acclaimed GPT-5.5, recognized as the company's most intelligent and intuitive model.

This restricted deployment approach reflects a growing trend in the AI industry, where companies choose to limit access to their most powerful models to prevent misuse. OpenAI has already adopted this strategy with other cybersecurity-focused models, as well as with GPT-Rosalind, a model dedicated to life sciences.

Meanwhile, Anthropic has recently taken a similar approach with its model Claude Mythos, although the launch has been marred by difficulties. The White House, despite tensions with Anthropic, has expressed concerns about broadening access to Mythos, fearing that it could compromise cybersecurity and the government's ability to effectively utilize the system.