Claude Mythos marks major AI leap in cybersecurity capabilities, released under restricted access through Project Glasswing

Claude Mythos represents a significant advancement in AI, particularly in cybersecurity, with enhanced ability to autonomously assemble complex exploits for major software

Model access limited to select cybersecurity firms through Project Glasswing initiative to allow patching of critical software vulnerabilities before exploitation

Article coverage included System Card documentation, cybersecurity capabilities analysis, and additional technical notes on the model

Separate security incident: physical attack attempted on OpenAI CEO Sam Altman, though the attempt was unsuccessful