Anthropic is expanding Project Glasswing to 150 new organizations. The initiative now operates across more than 15 countries. Total participation has reached approximately 200 organizations.
Selected partners receive access to Claude Mythos Preview, Anthropic's most capable and restricted AI model. The model scans for software vulnerabilities within critical infrastructure sectors including power, water, healthcare, and communications.
An initial phase involved 50 partners. The AI model identified over 10,000 high- or critical-severity security flaws during that period. Anthropic is broadening access to ensure safeguards exist before similar models are developed elsewhere.