Anthropic Faces US Export Control on Mythos Models, Plans ID Checks by July 8

Anthropic has suspended its Mythos 5 and Fable 5 models after the Trump administration issued an export control order on June 22, 2026, citing national security concerns. The order, which Bloomberg reported, restricts use by any foreign national, including Anthropic's own employees, effectively halting access for all customers.

The company received the letter from the Bureau of Industry and Security with a 90-minute deadline to comply, leaving no time for clarification. Anthropic responded by disabling both models for all users, a move that cybersecurity expert Katie Moussouris has criticized as unjustified.

Amazon Review Raises Questions

Anthropic had commissioned Amazon to review both models before releasing Fable 5. The resulting report, which Moussouris confirmed she saw, has been cited by the government to justify the order. Moussouris, known for helping create Microsoft's bug bounty program and leading the Hack the Pentagon effort, called the government's reasoning into question.

• The order treats Mythos and Fable as dual-use technology requiring export licenses.
• It blocks use by foreign nationals inside and outside the US, including Anthropic staff.
• Anthropic had no prior warning and received no detailed explanation from officials.
• Moussouris stated that the Amazon report does not support the sweeping restrictions imposed.

ID Verification Plans Advance

While navigating the export crisis, Anthropic announced it will require identity verification for certain capabilities starting July 8, 2026. The move, posted on Reddit and discussed on Hacker News, aims to restrict access to features that could be misused. The timeline suggests the company is tightening controls amid regulatory pressure, though it remains unclear which specific capabilities will be affected.

Industry observers note that open-weight models could soon match Mythos's capabilities, making such restrictions harder to enforce. The Trump administration's action may set a precedent for controlling advanced AI, but experts warn that broader deployment of open-source alternatives will challenge any such regime.

Anthropic has not commented on whether the ID verification applies globally or only in the US. The company faces a tangled path ahead: complying with export controls while managing user access and preparing for a future where similar models are freely available.