Anthropic is expanding Project Glasswing, a cybersecurity initiative that uses its Claude Mythos Preview model to help identify and address vulnerabilities, to an additional 150 organizations. Multiple outlets report that the expansion follows work with an initial cohort of roughly 50 partners and other stakeholders, including security firms, open-source maintainers, and government agencies. Anthropic says organizations must meet specified security requirements before receiving access. The expanded program now includes organizations in more than 15 countries. Outlets also describe Project Glasswing’s broader industry reach, with participants across sectors such as healthcare, energy, communications, technology, and other infrastructure. CNBC and others note that the effort was initially released in April as a limited access program to test the model’s ability to support cybersecurity flaw discovery and related activities. Infosecurity Magazine adds that the broader rollout is occurring as patching is increasingly seen as a bottleneck in vulnerability management. Across coverage, Anthropic frames the expansion as a step in scaling partner access while maintaining controls around eligibility and security.