Anthropic has restricted access to its powerful new Claude Mythos AI model to security researchers under Project Glasswing, citing its unprecedented cybersecurity capabilities. The model has already discovered thousands of high-severity vulnerabilities across major operating systems and browsers, including complex exploit chains. This controlled release aims to give the industry time to prepare defenses before such capabilities potentially become widely available.
Background
AI models are increasingly capable of identifying and exploiting software vulnerabilities, raising concerns about responsible deployment. Anthropic follows a pattern of cautious release similar to other AI safety initiatives.
- Source
- Simon Willison
- Published
- Apr 8, 2026 at 04:52 AM
- Score
- 8.0 / 10