Anthropic Quadruples Project Glasswing to ~200 Partners in 15+ Countries as Mythos Goes After Critical Infrastructure

Anthropic said on June 2 that it is quadrupling the size of Project Glasswing, its program for putting the unreleased Claude Mythos model in the hands of organizations that defend critical software. Roughly 150 new partners across more than 15 countries are joining the initiative, taking it from the ~50 organizations announced in early April to around 200 — with further expansion planned.

The new cohort deliberately reaches beyond Big Tech. Alongside existing partners like Apple, Nvidia, Microsoft, CrowdStrike and Palo Alto Networks, the expansion pulls in sectors that have been underrepresented so far: power, water, healthcare, communications and hardware, plus vendors maintaining critical open-source codebases. Several participating organizations estimate that a successful attack on their systems could affect more than 100 million people each.

Partners get access to a preview of Claude Mythos — the frontier model Anthropic has so far declined to release publicly — to scan codebases for vulnerabilities, write patches, run pre-release checks and penetration tests, power threat detection and response, and in some cases rebuild legacy code in memory-safe languages. The early results are striking: Glasswing partners have already surfaced more than 10,000 high- or critical-severity flaws, and organizations must meet Anthropic’s security requirements before getting access at all.

Anthropic is unusually candid about why it is racing to widen the program. "Cheap, fast AI models with powerful cyber capabilities are around the corner," the company said, adding that it expects Mythos-class competitors to emerge within 6 to 12 months. The expansion was coordinated with the security industry, open-source maintainers and the U.S. government — an attempt to give defenders a head start before equivalent offensive capability becomes broadly available.