Solutions  ·  2026-04-11

Anthropic Project Glasswing: Claude Mythos Preview Discovers Thousands of Zero-Days

SolutionsHigh impact
Anthropic launched Project Glasswing, providing Claude Mythos Preview — a model with exceptional security research capabilities — to 12 launch partners including AWS, Apple, Google, Microsoft, NVIDIA, CrowdStrike, and Palo Alto Networks. Mythos autonomously discovered thousands of zero-day vulnerabilities across Windows, Linux, macOS, Chrome, Firefox, and Safari, including a 17-year-old FreeBSD RCE (CVE-2026-4747). Anthropic is committing $100M in usage credits and $4M to open-source security organisations.
This represents a step-change in AI-assisted vulnerability discovery. The model's ability to autonomously find and exploit decades-old vulnerabilities across major platforms signals that both defenders and attackers will soon have access to similar capabilities, fundamentally altering the vulnerability disclosure landscape.
Critical infrastructure operators, OS vendors, browser vendors, and open-source maintainers should monitor Glasswing disclosures. Security teams should prepare for a significant increase in vulnerability disclosures in the coming weeks as partners run Mythos against their codebases.
Sources
Anthropic — Project GlasswingNPR — Anthropic Project Glasswing AI CybersecurityThe Hacker News — Claude Mythos Finds Thousands of Zero-Day FlawsTechCrunch — Anthropic Mythos AI Model Preview Security
See this in the live feed Explore related AI security and governance findings — updated every morning.
Open the feed →