U.S. Partly Lifts Anthropic Block, Allowing Claude Mythos 5 to Reach Approved American Users
The U.S. government on Friday lifted its block on Anthropic’s powerful AI model Claude Mythos 5, allowing the company to release it to more than 100 U.S. entities, including major companies and government agencies, according to Semafor. The decision, delivered in an afternoon letter to Anthropic, eases tensions between the Trump administration and one of the world’s most valuable private companies.
The administration had imposed export restrictions on Mythos two weeks earlier, which led Anthropic to shut down the model and also close its related model, Fable 5. Amazon and other companies had warned that the safety restrictions could be bypassed and the systems could be used for harmful purposes. Friday’s letter did not address Fable 5, a weaker version of Mythos that briefly became the most powerful widely available AI model for consumers.
In the letter to Anthropic computing chief Tom Brown, Commerce Secretary Howard Lutnick said he had determined that “appropriate safeguards” now allow certain trusted partners to access Claude Mythos 5. He said there had been “significant progress” in the intensive daily talks between the government and the company since the block took effect, and noted that Anthropic had pledged to work with the U.S. government on “protocols, standards and releases” of its models. People familiar with the talks said a release of Fable is also being prepared, but the timing remains unclear.
Under the new arrangement, no export, re-export or in-country transfer license will be required for Claude Mythos 5 for the entities listed in the letter’s annex and for their foreign-national employees, or for Anthropic’s own foreign-national employees. The move marks the start of a new regulatory regime giving Washington control over the release of advanced AI models. Commerce Department spokesman Benno Kass said the government acted quickly, adding, “For just two weeks, we worked decisively to ensure America remains the global leader in artificial intelligence, while preserving our security.” The policy arrives as OpenAI launched GPT-5.6 the same day to a limited list of government-approved partners, and European and other allies have voiced frustration at their new dependence on Washington’s decisions. U.S. officials had also worried that Mythos might be released to a partner with too-close ties to China, reportedly a South Korean telecom provider.
