The US Commerce Department on Friday removed export restrictions on Anthropic's Claude Mythos 5 AI model, permitting distribution to over 100 US institutions including major corporations and government agencies. This marks a significant pullback in the Trump administration's tensions with one of the world's most valuable private AI companies.

Two weeks earlier, the administration had imposed export controls on both Mythos and its lighter-weight variant Fable 5, citing concerns that Amazon and other companies raised about potential security vulnerabilities. Officials worried these models could be compromised or misused for malicious purposes. Commerce Secretary Howard Lutnick's letter indicates that intensive daily negotiations between the government and Anthropic produced sufficient safeguards to justify the policy shift.

The decision arrives alongside OpenAI's release of GPT-5.6 to government-approved partners, suggesting coordinated regulatory movement across major AI players. Under the new arrangement, Anthropic can provide Mythos 5 access to approved entities and their foreign employees without requiring individual export licenses, streamlining deployment for trusted partners.

This resolution establishes precedent for how US authorities will balance national security concerns against AI innovation, particularly as companies seek broader distribution while regulators grapple with preventing adversarial exploitation of advanced models.