Anthropic’s Claude Mythos is now available, but not for you

Anthropic has limited access to its new Claude Mythos Preview model to a small set of partners for cybersecurity testing.

Under Project Glasswing, the model is helping uncover and fix vulnerabilities in enterprise systems and open-source software.

Anthropic says it has already found thousands of zero-day issues, but is delaying public release because of security risks.

The move underscores how advanced AI is being tightly gated to support defenders without increasing attacker capabilities.