US lifts block on Anthropic's Claude Mythos 5 for 100+ trusted organizations

The US government on Friday lifted its export controls on Anthropic's Claude Mythos 5 AI model, granting over 100 trusted US institutions—including major companies and government agencies—access to one of the world's most advanced AI systems. Commerce Secretary Howard Lutnick communicated the decision in a letter to Anthropic's chief compute officer, citing 'significant progress' in intensive daily talks since the block was imposed two weeks ago. The initial restriction followed warnings from Amazon and others that Mythos could be 'jailbroken' for malicious purposes, and concerns over its release to partners with potential ties to China. Under the new arrangement, a license is no longer required for transfers to entities listed in Annex A and their foreign national employees. The weaker Fable 5 model remains blocked, though sources indicate talks are moving toward its release as well.

The decision marks a major de-escalation between the Trump Administration and Anthropic, one of the world's most valuable private AI companies. It also establishes a new regulatory precedent, giving the US government significant control over the release of frontier AI models. The framework is being built on the fly, leaving many international partners—including European allies and non-US companies—uncertain about when they will gain access. The same day, OpenAI released its GPT-5.6 model to a short list of government-approved partners, highlighting the emerging pattern of selective, security-constrained AI distribution. Anthropic has committed to working with the US government on protocols and standards for future releases, as the global AI race intensifies.