Project Glasswing: Anthropic says Claude found 10,000 critical software flaws in a month

TL;DR AI
2 min readKey summary
Anthropic says its Project Glasswing program used Claude Mythos Preview to uncover more than 10,000 high- and critical-severity vulnerabilities in one month.
Some of the findings were independently confirmed, and at least one major open-source cryptography flaw was patched, including CVE-2026-5194.
The report underscores how AI is speeding up bug discovery faster than organizations can fix issues, creating a new security bottleneck.
Cloudflare, Mozilla, Firefox, wolfSSL, Palo Alto Networks, Microsoft, and Oracle were among the systems and projects affected or examined.
