Anthropic launched Project Glasswing on April 13, 2026, a defensive cybersecurity initiative powered by the Claude Mythos Preview AI model. The model identified thousands of critical security vulnerabilities across all major operating systems and web browsers. This discovery signals a shift toward high-speed, large-scale AI vulnerability research.
U.S. intelligence and defense officials expressed alarm over the potential for adversaries to weaponize the tool. National security circles are debating how to remediate infrastructure weaknesses without exposing actionable data to hostile actors.
Access remains restricted to a select group of partners including Amazon Web Services, Apple, Cisco, Google, and Microsoft. These companies face an immediate challenge to patch the newly discovered flaws.