OpenAI: Hey, We Also Have a New Tool That Is So Scarily Powerful We Can’t Release It

OpenAI is reportedly preparing a powerful new cybersecurity tool that it considers too risky to release broadly, echoing Anthropic’s recent unveiling of its Mythos model. According to Axios, the tool will be offered only to select partners and is not a new model but a specialized product designed to enhance defensive security work. OpenAI already runs a limited “Trusted Access for Cyber” program, but the company now appears eager to match Anthropic’s high‑profile claims and maintain momentum in the AI security race. Despite the secrecy, OpenAI continues projecting massive future revenue growth, even as its current ad business remains modest.

Read the full story on Gizmodo →