White House demands OpenAI and Anthropic block UK access to new AI models

Full text

The White House has asked OpenAI and Anthropic not to share their new artificial intelligence models with the U.K. government’s testing agency until the models have gone through testing with the U.S. government, according to a person familiar with the matter and a senior U.S. administration official, both of whom were granted anonymity to describe the request.

The request, which came from the Office of the National Cyber Director, puts companies in the difficult position of having to choose between withholding their models from U.K.’s AI Security Institute – one of the most well-resourced government testing agencies in the world which previously enjoyed privileged early access to models – or risk upsetting the Trump White House. The senior administration official said the White House wants the U.S. to review the models and take steps to make sure U.S. systems are secure before the models are shared with U.S. partners.

“Because they’re American companies and this has been our policy with every new frontier model that comes out,” the senior official explained.

The White House did not immediately respond to a request for comment.

Anthropic appears to have agreed to the request, so far. The company did not provide its Claude Mythos 5.1 model to U.K. AISI, saying in its announcement that the model was “only available to a set of U.S. organizations.”

“We’re coordinating with the U.S. government to expand access to a broader set of domestic and international partners as quickly as possible,” an Anthropic blog announcing the model reads. Anthropic declined to comment for this piece.

In a letter to a U.K. parliamentary committee earlier this month, AISI Director Henry de Zoete acknowledged the lack of access to Anthropic’s model, but said: “We maintain strong relationships with all frontier AI developers and continue to have prerelease access to some of the world’s most capable models,” pointing out that it tested OpenAI’s GPT-6 Astra model ahead of release.

“The trusted relationships AISI holds enable us to test a range of models that represent industry-wide capability jumps,” de Zoete wrote.

OpenAI did not immediately respond to a request for comment.

The Trump administration’s request comes as the White House tries to determine how the government should handle powerful, new AI models that have hacked into outside entities during testing – including, as revealed yesterday, an Australian government website .

It also comes as former engineers and leading officials from major AI companies warn of possible catastrophe if their models are left to advance unchecked.

Created in 2023, the U.K.’s AISI has established close links with leading AI labs, routinely testing their latest models ahead of public release on a voluntary basis.

The U.K. government has made AISI’s expertise and model access a central part of its domestic response to AI safety concerns, as well as its pitch to lead work on a global agreement on AI.

Prime Minister Andy Burnham this week said AI will be “at the heart” of the country’s G20 presidency next year.

But AISI’s privileged access to frontier models from leading American labs is under pressure from the Trump administration’s insistence that the U.S. government get first access to emerging systems.

The U.K. has sought to downplay any clash, with Burnham saying at the U.N. General Assembly this week that the U.K.’s institute was “working hand in glove with the U.S. and the leading labs on AI safety.” He called for a “single set of global principles and standards” to improve transparency and preparedness and ensure AI is developed safely.

U.K. Foreign Secretary Ed Miliband made a similar case at the U.N. Security Council on Wednesday, saying governments need to ensure frontier models are “rigorously tested” and have enough visibility to assess what companies are doing.

“The leading AI companies have actually committed to provide this visibility. It’s really, really important, and it is an offer that we and they should follow through,” Miliband said.

“The U.K. is a world leader in AI security, and the AI Security Institute continues to work closely with the U.S. government, international partners and leading AI companies, including OpenAI and Anthropic,” a U.K. government spokesperson said.

“These risks do not stop at national borders and no country can tackle them alone. We will continue to work closely with the U.S. and other partners on the testing of advanced AI, while building our own rigorous scientific understanding of these systems and ensuring our approach is grounded in the evidence.”

In the U.S., Congress and the White House are pushing for more resources for the Commerce Department’s Center for AI Standards and Innovation, or CAISI. It’s seen as the crown jewel of AI governance in the U.S. but it is currently operating without a permanent director and is increasingly facing a bandwidth problem as it is also asked to evaluate an expanding stream of frontier models while operating with only a few dozen technical employees.

OpenAI said this week that CAISI should lead work on international standards to monitor AI progress and safety, working with other national AI safety institutes including the U.K.’s AISI.