WASHINGTON – US Commerce Secretary Howard Lutnick warned Anthropic in a letter that it would need government permission to grant foreign nationals access to its most advanced AI models and threatened criminal and civil penalties if the firm failed to comply, according to a copy seen by Bloomberg News.
The letter, dated June 12, ordered Anthropic not to give its Fable 5 and Mythos 5 artificial intelligence models to foreign nationals anywhere in the world without a licence from the US Commerce Department. Lutnick gave no basis for why the restrictions were necessary, but his letter cited US laws that allow the government to impose export controls on civilian technology that could be used for intelligence purposes by an adversary’s military.
“Until further notice, you must submit an application for an individually-validated license prior to the export, re-export, or transfer (in-country), including deemed export or deemed re-export, of the Mythos or Fable models to any destination worldwide or to any ‘foreign person’ wherever located,” Lutnick wrote in the letter, addressed to Anthropic Chief Executive Officer Dario Amodei.
A Commerce Department spokesperson declined to comment, while an Anthropic spokesperson directed Bloomberg to the company’s blog post on the export controls.
The Commerce Department directive prompted Anthropic late on June 12 to disable all access to the two models. Since then, representatives from Anthropic have held virtual meetings with US officials about specific security issues, and technical staff from the AI start-up met with US officials at Commerce on June 15, a company spokesperson said.
While the Trump administration has not commented on the reasons for its decision, the company has said it believes the US government issued the order after discovering that it’s possible to “jailbreak”, or bypass the guardrails, of Fable 5, a recently released version of Mythos that Anthropic had blocked from carrying out cybersecurity tasks.
The order from Lutnick represents the most significant intervention by the US government to date into an AI venture’s operations. It poses a new challenge to Anthropic weeks after the company filed confidentially for an initial public offering, with its latest valuation topping US$900 billion (S$1.15 trillion).
In a statement announcing that it had disabled access to the two models, Anthropic said it regarded the government move as disproportionate.
“We disagree that the finding of a narrow potential jailbreak should be cause for recalling a commercial model deployed to hundreds of millions of people,” Anthropic said in its post. “If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.”
Lutnick, who is travelling with US President Donald Trump for the Group of Seven summit in France, said in his letter that the licence requirement would remain in effect until further notice and spelt out the procedures for Anthropic to follow in submitting an application. BLOOMBERG