Anthropic’s Mythos AI Uncovered Serious Security Holes in Every Major OS and Browser

Anthropic has developed a new AI model called Mythos with unprecedented cybersecurity capabilities, able to autonomously discover and exploit zero-day vulnerabilities in every major operating system and web browser. The model dramatically outperforms its predecessors — finding 181 Firefox JavaScript engine exploits versus just 2 for the previous best model — and can chain vulnerabilities into complex attacks. Due to its potency, Anthropic is not releasing Mythos publicly, instead sharing it with select tech companies and open source developers via Project Glasswing to help patch vulnerabilities before malicious actors gain similar capabilities. Anthropic warns that traditional 'defense-in-depth' security measures may become ineffective against AI-assisted attackers, and expects competing labs to produce models with similar capabilities within 6–18 months.