Anthropic Unveils Powerful Mythos Model for Cybersecurity

Anthropic has launched Project Glasswing, a cybersecurity initiative deploying its most powerful AI model, Claude Mythos, to over 40 partner organizations including Amazon, Apple, Microsoft, and Cisco. The company is withholding public release due to safety concerns, instead committing $100 million in API credits to support defensive security research. Mythos has already identified thousands of vulnerabilities across major operating systems and browsers, including a 16-year-old FFmpeg bug missed by traditional tools over 5 million scans. The model scores 93.9% on SWE-bench Verified, far surpassing existing Claude Opus models, but its ability to instantly generate exploits makes controlled deployment essential.