Axios Scoop: OpenAI Building Cybersecurity Product to Rival Anthropic's Mythos — Trusted Access Pilot

Axios·high signal

OpenAI is finalizing a cybersecurity product with advanced hacking and defensive capabilities for release to a small set of partners through its existing 'Trusted Access for Cyber' pilot program, per Axios on April 9. The program launched in February after GPT-5.3-Codex shipped. This directly responds to Anthropic's Mythos Preview, which found 'tens of thousands of vulnerabilities' and triggered Project Glasswing's restricted-access coalition.