Anthropic has demonstrated a breakthrough in proactive cybersecurity. Its Project Glasswing initiative, launched April 7, 2026, deployed the Claude Mythos Preview model to scan global software. Within six weeks, it identified over 10,000 high- and critical-severity vulnerabilities.
Many are zero-days, previously unknown. Among them: a 27-year-old flaw in OpenBSD and a 16-year-old flaw in FFmpeg, the video framework underpinning countless applications. The AI didn't just spot isolated bugs; it chained them into working exploits, a task typically requiring elite human researchers.
Access to Claude Mythos Preview is restricted to vetted partners, including AWS, Apple, Google, and Microsoft, to patch critical infrastructure before attackers can exploit it.
The project highlights a looming crisis: AI now finds vulnerabilities far faster than humans can fix them. With a single model generating thousands of critical findings in weeks, the traditional remediation pipeline is becoming a bottleneck.