Trump Administration Limits Foreign Access to Anthropic's Mythos 5 AI Model

The Trump administration's recent export controls on Anthropic's AI model, Mythos 5, led to a swift suspension of access for foreign nationals, just days after its launch on June 9-10. The government cited national security concerns, prompting Anthropic to disable both Mythos 5 and its companion model, Claude Fable 5, on June 12. This drastic measure effectively cut off access for all users, not just those targeted by the Commerce Department's directive.

The ensuing two weeks were marked by intense negotiations between Anthropic executives and Commerce Secretary Howard Lutnick. The export controls required licenses for foreign nationals working in the U.S. to access the models, complicating matters for cybersecurity teams that included engineers on work visas. High-level discussions between Lutnick and Anthropic leadership were crucial in resolving the standoff.

To break the impasse, Anthropic implemented additional safeguards, although the specifics remain undisclosed. By late June, the Commerce Department agreed to lift the restrictions, but access was not restored to the general public or all paying customers. Instead, a limited number of users, including select U.S. companies, federal agencies, and partners involved in Project Glasswing, were granted access.

Mythos 5 is priced at $10 per million input tokens and $50 per million output tokens, with its primary applications focused on cybersecurity testing and scientific hypothesis generation. The resolution also allowed foreign national employees at approved companies participating in Project Glasswing to regain access, addressing a significant barrier that had rendered the models unusable for many organizations during the restrictions.