U.S. — Parts of the National Security Agency (NSA) have lost access to Anthropic's Mythos 5 artificial intelligence model. The Trump administration restricted the export of the model this month due to national-security concerns. Some NSA analysts were notified on Friday they would lose access to the Mythos model.

Following the implementation of these export controls, Anthropic pulled back the release of its most advanced models, including Mythos 5 and Fable 5. The NSA's red teams, which conduct cybersecurity testing, no longer have access to Mythos. The red teams' authority to use Mythos came through Project Glasswing, which launched in April. Project Glasswing is an effort to give select security researchers and organizations early access to Mythos Preview. Anthropic expanded the program to approximately 150 organizations across more than 15 countries. The NSA may still be able to use earlier versions of the technology under prior arrangements.

Virginia Senator Mark Warner, the top Democrat on the Senate Intelligence Committee, said NSA Director General Joshua Rudd told him that Mythos "broke into almost all of our classified systems, not in weeks, but in hours." An official from the U.S. stated that the specific Mythos work was part of a red-teaming effort designed to test the security of internal networks. The Economist's defense editor stated on X that a U.S. official told him Senator Warner had misunderstood General Rudd's comments. On Monday, the Five Eyes intelligence alliance issued a warning that frontier AI models could sharply change the cyber threat landscape within months.