US Allows Anthropic to Release Mythos to Trusted Organizations

The U.S. government has partially reversed its 12th June order blocking access to Anthropic's most advanced AI models, allowing the company to release its Claude Mythos 5 system to a select group of more than 100 American companies and institutions.

The approved organizations include many Fortune 500 firms that operate and defend critical infrastructure, a source familiar with the directive said. Anthropic had abruptly disabled both Mythos 5 and the consumer-facing Fable 5 two weeks ago after the Commerce Department imposed export controls, citing the risk that foreign adversaries could use the models to carry out sophisticated cyberattacks.

Commerce Secretary Howard Lutnick wrote to Anthropic that there had been significant progress in the company's work with the government to address the risks. The letter said an export license is no longer required for Mythos 5 to reach trusted companies and their non-U.S. citizen employees, though licensing restrictions remain for organizations not on the approved list. The government is also moving toward permitting a wider release of Fable 5, though a timeline remains unclear.

The arrangement has drawn sharp criticism. John Coleman, legislative counsel for the Foundation for Individual Rights and Expression, said the government's selection process lacks transparency and places too much power in the hands of the state. OpenAI chief executive Sam Altman, whose company is also delaying a full public launch of GPT-5.6 at the government's request, wrote on X that safety testing is reasonable but added, "I just don't like the idea of the government picking the customers".

Anthropic had earlier refused to allow the U.S. military to use its models for domestic surveillance and fully autonomous weapons, which led the government to place it on a national security blacklist. The current restrictions follow an executive order signed by President Donald Trump this month establishing a voluntary framework for AI developers to share frontier models with the government before release.

Kate Koren, an analyst at the Center for Strategic and International Studies, described the partial release as a practical interim step but warned that the longer a system is not in place for companies to widely release new models, the more likely it is that China will catch up.