Anthropic Says Mythos Has Already Found More Than 10,000 Vulnerabilities

TL;DR AI
2 min readKey summary
Anthropic says Project Glasswing, powered by Claude Mythos Preview, helped partners find more than 10,000 vulnerabilities in about a month.
The findings included many high- and critical-severity bugs in products from companies such as Cloudflare, Mozilla, and Microsoft.
Anthropic has not released the model publicly, saying its safeguards are not yet strong enough.
The company plans to expand access later through partners and governments, highlighting AI’s growing role in cybersecurity.
