Anthropic's Claude Mythos Identifies 271 Security Vulnerabilities in Firefox

Gate News message, April 22 — Anthropic's latest AI model, Claude Mythos, identified 271 security vulnerabilities in Mozilla Firefox during internal testing, with all flaws remedied this week, Mozilla disclosed. In contrast, a previous version of Anthropic's model had detected only 22 security-sensitive vulnerabilities.

Mozilla stated that all discovered vulnerabilities fell within the scope of findings by top human security researchers. Claude Mythos, launched in March 2026, is Anthropic's most powerful model for reasoning, coding, and cybersecurity applications. The model is currently available to vetted partners including Amazon, Apple, and Microsoft through the "Project Glasswing" program.