Anthropic Opens Mythos to 150 New Partners Across Critical Infrastructure Sectors

Anthropic has expanded its Project Glasswing initiative, granting 150 additional organizations across more than 15 countries access to its Mythos model for software vulnerability testing. The move, announced Tuesday, brings in partners from industries underrepresented in the initial rollout—including energy, water utilities, healthcare, telecom, and hardware manufacturing. New entrants must pass security vetting before using the model. “This is a deliberate step toward making AI a standard tool for software security, and helping the industry rethink core cybersecurity assumptions,” the company stated in a blog post. The expansion follows Anthropic’s Monday announcement that Mythos would be available in the European Union, and the confidential filing of its IPO prospectus with the SEC—beating rival OpenAI to that milestone. The company first tested Mythos with 50 partners in April, amid concerns that hackers could weaponize the model to accelerate vulnerability discovery. Since then, Project Glasswing participants have identified over 10,000 high or critical security flaws. Anthropic estimates a major cyberattack exploiting such vulnerabilities could affect more than 100 million people. Existing major partners include Apple, Nvidia, Microsoft, CrowdStrike, and Palo Alto Networks. Anthropic has not named the new coalition members.