Read, Watch or Listen

Washington, United States – The U.S. National Security Agency has lost access to Anthropic’s advanced artificial intelligence model, Mythos 5, after the system rapidly uncovered serious software vulnerabilities in highly classified government networks during a test under a restricted program known as Project Glasswing. NSA Chief Joshua Rudd told a congressional hearing this month that the model effectively broke into almost all of the agency’s classified systems in hours rather than weeks, exposing weaknesses across highly secure government networks and intensifying concern in Washington over the risks and benefits of deploying powerful commercial AI in national security systems. The loss of access comes amid an escalating dispute between Anthropic and the Trump administration, which has ordered the company to halt exports of its latest Mythos and Fable models worldwide and to all foreign nationals, citing national security concerns. In February, the Department of Defense designated Anthropic a “supply chain risk” after the firm refused to lift restrictions on certain military applications, including widespread domestic surveillance and fully autonomous weapons. President Donald Trump subsequently directed federal agencies to phase out Anthropic’s technology within six months, and the company has sued the U.S. government, alleging unlawful retaliation over its limits on military use of AI. The New York Times first reported the NSA’s loss of access to Mythos, and the White House and Anthropic have not responded to requests for comment, as the company’s AI tools remain integrated into U.S. intelligence analysis, operational planning, and cyber operations while it prepares for a potential initial public offering.

Prepared by Jonathan Pierce and reviewed by editorial team.