Anthropic's Claude Mythos finds over 10,000 critical software flaws

Technology May 23, 2026

Anthropic's new AI, Claude Mythos, found over 10,000 critical software flaws in only one month as part of Project Glasswing. The goal? To make digital infrastructure safer for everyone.
Big names like Cloudflare and Mozilla are already seeing better bug detection thanks to this tech.

Mythos stops $1.5 million bank fraud attempt

Cloudflare uncovered 2,000 bugs (400 high-risk) with fewer false alarms than humans.
Mozilla caught 271 vulnerabilities in Firefox 150, over 10 times more than before.
Mythos also scanned open-source projects and stopped a $1.5 million bank fraud attempt.
Anthropic admits fixing these bugs fast is tough and puts extra pressure on cybersecurity teams, but they're working on it.