OpenAI rolls out new model for cybersecurity teams a month after Anthropic's Mythos debut

On Thursday, OpenAI introduced GPT-5.5-Cyber, a specialized version of its latest artificial intelligence model, which is now available in a limited preview for select cybersecurity teams. This announcement comes just a month after Anthropic made waves in the tech community with the launch of its Claude Mythos Preview, attracting significant attention from investors and government officials alike. While OpenAI clarified that GPT-5.5-Cyber does not represent a drastic upgrade in cybersecurity capabilities, it has been specifically trained to handle security-related tasks more flexibly. This tailored approach aims to assist cybersecurity teams in crucial workflows such as identifying and triaging vulnerabilities, validating patches, and analyzing malware, which could be complicated by the safeguards present in the standard GPT-5.5 model. In its announcement, OpenAI emphasized that this model allows a select group of partners to explore advanced workflows where specialized access behaviors can be particularly beneficial. This strategic move echoes Anthropic's recent decision to restrict access to its Mythos model to a limited number of companies as part of its cybersecurity initiative, Project Glasswing. Anthropic's CEO, Dario Amodei, has been active in discussions with high-ranking officials in the Trump administration regarding the capabilities of Mythos, even after the company faced restrictions from the Pentagon. Notable figures, including Federal Reserve Chairman Jerome Powell and Treasury Secretary Scott Bessent, have engaged with major U.S. bank leaders to discuss the implications of Mythos, further highlighting the growing intersection of AI and cybersecurity in today's landscape.