Senior technical staff from Anthropic are in Washington this week meeting with White House officials after the Trump administration ordered the company Friday to cut off foreign nationals from its two most advanced artificial intelligence models, sources confirmed Monday.
The meetings, reported by The Hill, follow a rapidly escalating standoff between the San Francisco AI firm and the administration over national security oversight of frontier AI. A source close to Anthropic confirmed the company’s top staff have been holding daily virtual meetings with federal officials since the White House first contacted them.
The administration’s export control directive, issued Friday afternoon, ordered Anthropic to immediately suspend access to its Fable 5 and Mythos 5 models for any foreign national. Anthropic complied and disabled both models the same day. The company had released the models just four days earlier, on June 10.
According to Anthropic, federal officials cited national security concerns tied to a potential method of circumventing the models’ built-in safeguards. The technique, the government said, essentially involved directing the AI to read a specific software codebase and identify security flaws — an approach that critics warn could supercharge hacking operations in adversarial hands.
Anthropic disputed the severity of the threat in a statement issued Friday. “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,” the company wrote. “If this standard was applied across the industry, we believe it would essentially halt all new model deployments for all frontier model providers.”
The company characterized the jailbreak as “non-universal,” meaning it could not broadly override the model’s protections. Anthropic acknowledged that “perfect jailbreak resistance is not currently possible” and said it stands by its current safety approach.
Anthropic also said the administration’s directive did not conform to principles the company has publicly backed, including a statutory process it described as “transparent, fair, clear, and grounded in technical facts.” The company said it supports government authority to block unsafe AI deployments, but said Friday’s order fell short of that standard.
The Fable 5 model was made available to the general public last week with safety restrictions in place. Mythos 5, which has some of those restrictions removed, was provided only to a select group of cyberdefense firms and critical infrastructure providers. Anthropic had previously declined to release a version of Mythos more broadly over concerns it could strengthen offensive hacking capabilities.
The administration’s move marks the first time the federal government has used export control authority to pull a major commercial AI model that was already in public release. Officials did not provide Anthropic with the specific intelligence underlying the national security finding, according to the company.
White House officials have not commented publicly on the ongoing talks with Anthropic. The company said it is “committed to working constructively with the government” to resolve the matter.
