NSA loses access to Anthropic’s Mythos 5 model after U.S. export controls

Parts of the National Security Agency have lost access to Anthropic’s Mythos 5 model after the Trump administration imposed export controls on the company earlier in June 2026, according to two people familiar with the matter. The agency may still retain limited access to earlier versions of the model under prior arrangements, even as company support, updates, or modifications are now more restricted.

The access issues stem from the administration’s decision to restrict Anthropic’s release of its most advanced models, including Mythos 5 and Fable 5, citing national security concerns. The move forced Anthropic to pull back the rollout of these models, disrupting at least some NSA work involving one of the most closely watched AI tools in government.

NSA’s red-team efforts involving Mythos were conducted under Project Glasswing, a program launched in April 2026 to provide select security researchers and organizations with early access to Mythos Preview. The program was later expanded to roughly 150 organizations across more than 15 countries, including critical infrastructure operators and cyber defenders, following collaboration with government, industry, and open-source software partners.

The loss of access to Mythos 5 comes as the Five Eyes intelligence alliance warned that frontier AI models could sharply change the cyber threat landscape within months, not years, by accelerating both cyberattacks and cyber defense. The alliance’s warning underscores the broader implications of restricted access to advanced AI tools for national security workflows.