Expanding Project Glasswing (www.anthropic.com)

Project Glasswing is expanding its initiative to enhance cybersecurity by increasing access to its Claude Mythos Preview model. Originally involving 50 partners who identified over 10,000 security vulnerabilities in their codebases, the initiative is now growing to approximately 150 new organizations across more than 15 countries, including critical sectors like power, healthcare, and communications. This expansion is vital as many partners maintain codebases on which millions rely, meaning any breach could have severe implications for global security. The project aims not only to find vulnerabilities but also to streamline the processes of disclosing, verifying, and patching them. Claude Mythos Preview is being utilized for tasks such as writing patches and conducting penetration testing, enhancing the industry’s defense against cyber threats. Moving forward, Project Glasswing intends to develop new standards and infrastructure for cybersecurity while exploring safer distribution of powerful AI tools, all amid growing concerns that powerful AI models may soon be misused in cyberattacks. The initiative underscores the necessity for organizations to adapt to an evolving landscape where AI capabilities continually influence cybersecurity practices.