US Commerce Secretary Howard Lutnick warned Anthropic that it
must obtain government permission before giving foreign nationals access
to its most advanced AI models, according to a Bloomberg report citing a letter sent to CEO Dario Amodei.
The
letter, dated Friday, ordered Anthropic not to provide its Fable 5 and
Mythos 5 models to foreign nationals anywhere in the world without a
license from the Commerce Department.
Lutnick cited
US export control laws covering civilian technology that could be used
for intelligence purposes by an adversary’s military, though the letter
did not explain why the restrictions were necessary.
The
directive prompted Anthropic to disable access to both models late
Friday. The company has since held talks with US officials, including
meetings between technical staff and Commerce Department officials, as
it seeks to resolve the issue.
Anthropic has said it
believes the order was issued after the US government became aware of a
possible jailbreak in Fable 5, a recently released version of Mythos
that the company had restricted from performing cybersecurity tasks.
The
company called the move disproportionate and warned that applying the
same standard across the industry could halt new frontier model
deployments.
The order represents one of the most
aggressive US government interventions yet into the operations of a
frontier AI company. It comes weeks after Anthropic confidentially filed
for an IPO, with its latest valuation reportedly above $900 billion.
The
restrictions also raise broader questions for the AI industry. If
advanced models can be pulled from global markets through export
controls, AI labs may face new uncertainty around international revenue,
enterprise contracts, and access for foreign employees or customers.
Lutnick
said the license requirement would remain in effect until further
notice. The Commerce Department declined to comment on the specifics of
the letter, while Anthropic pointed to its public statement on the
export controls.