Anthropic debuts preview of powerful new AI model Mythos in new cybersecurity initiative

Anthropic announced a limited preview of its frontier model Mythos as part of Project Glasswing with 12 partner organizations to perform defensive security scanning; the company claims the model identified thousands of zero-day vulnerabilities, many critical. The report also details earlier security lapses tied to Anthropic—an exposed draft blog and an accidental exposure of nearly 2,000 source code files and subsequent GitHub repository takedowns—and notes the Mythos preview will remain limited to select organizations.