Anthropic scales Claude Mythos to critical infrastructure in 15 countries (techcrunch.com)

Anthropic is scaling its Project Glasswing initiative, which aims to identify and rectify critical software vulnerabilities using AI, to about 150 new organizations across over 15 countries. This expansion centers around Claude Mythos, Anthropic's latest and most powerful model, capable of detecting thousands of zero-day vulnerabilities within weeks. Initially provided to 50 partners, including the U.S. government, the technology is now further accessible to sectors such as power, water, healthcare, communications, and hardware, which are vital for global and national security. The organizations involved often maintain codebases that could impact over 100 million people in the event of a successful cyber attack. The widespread adoption of Claude Mythos is significant as it not only enhances security measures for critical infrastructure but also signals a competitive race in the AI space, particularly in cybersecurity. Anthropic's commitment to this initiative comes in response to the growing threat landscape, as competitors like OpenAI introduce their own models, such as GPT-5.5-Cyber, aimed at similar vulnerabilities. With partnerships extending to notable entities like Okta, Samsung, NATO, and EU cybersecurity agency ENISA, the initiative underscores the pressing need for robust AI-driven defenses in protecting digital assets and infrastructure against sophisticated cyber threats.