Pentagon Deploys Anthropic’s Mythos AI for Cybersecurity Amid Planned Breakup

The US Pentagon is currently deploying Anthropic’s Mythos artificial intelligence model to identify and patch critical software vulnerabilities across government systems, despite ongoing efforts to terminate its relationship with the AI firm. Emil Michael, the Defense Department’s chief technology officer, described this deployment as a necessary national security measure due to Mythos's unique ability to detect decades-old flaws in web browsers and infrastructure rapidly. This initiative falls under Anthropic’s Project Glasswing, which allows select organizations to use the unreleased Claude Mythos Preview for defensive cybersecurity purposes. The Department of Defense previously designated Anthropic as a supply-chain risk after failing to reach an agreement on model usage guidelines, prompting Anthropic to sue the administration in March to reverse the blacklisting. Michael noted that while Anthropic currently holds a temporary advantage in vulnerability detection, competitors like OpenAI, xAI, and Google are expected to release similar capabilities soon. The situation highlights the tension between immediate cybersecurity needs and long-term supply chain security concerns within the US defense sector.