Limited Return of Anthropic’s Mythos: What Happened

Anthropic, the U.S. company behind the Claude family of A.I. systems, developed a powerful model named Mythos. On June 12 the Department of Commerce directed Anthropic to suspend access to Mythos and Fable 5 for foreign nationals, and Anthropic disabled those models worldwide because it could not practically enforce nationality-based limits. The suspension interrupted planned product access and sparked legal and policy discussion.

Testing and reviews showed security concerns: the Associated Press reported that Mythos identified vulnerabilities in classified U.S. government systems during a red-team exercise. That finding helped drive urgent government review, and it was one reason officials sought stricter controls while they assessed the model's risks.

On June 26 the Commerce Department issued a letter that allowed Anthropic to return Mythos in a limited way to some trusted U.S. partners, subject to licensing and further evaluation. Although the release is narrow, it has been framed by officials as a step that eases tension between the company and the administration, while the broader export controls and licensing processes remain unresolved. The arrangement is limited to U.S.-based partners and does not restore full public access.

Lawmakers and industry groups have asked for more detail on the Commerce Department's authority. If companies and regulators continue to cooperate, the sector may develop clearer rules for frontier models; nevertheless, firms will need to balance rapid innovation with measures that reduce security vulnerabilities, and observers expect additional guidance in the coming weeks. Investors and researchers are monitoring whether the controls will affect product development and company financing.