The Trump administration is declining to make an exception for G7 allies in its new restrictions on Anthropic’s most advanced AI systems, even after British officials sought relief, according to people familiar with the matter.
The dispute centers on recent export controls that bar access to Anthropic’s top-tier models outside the United States and to foreign nationals. After the rules were announced last week on national security grounds, UK Prime Minister Keir Starmer asked for a carve-out that would allow British citizens and companies to use the company’s most advanced tools, including models identified by Anthropic as Mythos and Fable.
A Trump administration official said granting such an exemption to any other country, including a close ally, would not make sense. The official said the administration is working with Anthropic to ensure the technology is safe for users around the world.
White House officials are still discussing the issue with Anthropic, and one senior official suggested the matter could be resolved directly with the company’s chief executive, Dario Amodei. Another source described the talks as focused on preventing advanced models from being used in ways that could create security risks.
The administration’s position comes after Anthropic said the Commerce Department ordered it to block access to its most capable models for users outside the US and for foreign nationals. In response, the company disabled Fable 5 and Mythos 5 worldwide, saying it could not limit the models only to American users.
Anthropic has said the government’s concern involved a possible jailbreak that might allow the systems to identify software vulnerabilities. The company called that risk narrow and said it disagreed with the breadth of the response.
“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 after the restrictions were imposed.
The action came only days after the company introduced Fable 5, its latest high-capability model. Anthropic said it had worked with the government on safety measures before release.
The clash is the latest sign of a strained relationship between Anthropic and the Trump administration. Earlier this year, the government placed the company on a supply chain blacklist after Anthropic declined to allow the US military to use its models for domestic surveillance and autonomous weapons, according to the source material.
Anthropic has warned that if similar rules were applied across the AI industry, they could slow or effectively block the release of new frontier models by major developers. The company said its other models remain available.
The administration’s refusal to grant a G7 exemption underscores how seriously it is treating concerns about advanced AI access, even among allies. It also highlights the challenge of balancing US national security policy with the global reach of American AI companies and the expectations of foreign governments that rely on them.
The issue remains unresolved, and officials are continuing talks with Anthropic as pressure builds over whether the restrictions will stay in place or be modified through a negotiated agreement.