Anthropic's Claude Mythos Finds Thousands of Zero-Day Flaws Across Major Systems

Anthropic has developed Claude Mythos, a frontier AI model that demonstrated the ability to find and exploit thousands of high-severity zero-day vulnerabilities across major operating systems and web browsers. The model chained together four vulnerabilities to escape browser and OS sandboxes and solved a corporate network attack simulation that would take a human expert over 10 hours. Due to its advanced cybersecurity capabilities and concerns about potential misuse, Anthropic is not making the model generally available, instead using it in a controlled 'Project Glasswing' initiative with a limited set of organizations to secure critical software.