The U.K. won back access to pre-release OpenAI and Anthropic models

The U.K.'s AI Security Institute will resume receiving access to the most powerful AI models from OpenAI and Anthropic before their public release, AI Minister Kanishka Narayan announced on Monday, reversing a U.S. policy that had cut off such access.

Narayan credited lobbying efforts and recent conversations between ministers and tech executives — facilitated in part by a summit with King Charles III — for securing the change, according to Bloomberg. The Telegraph first reported the policy reversal, according to LBC.

The Trump administration had told OpenAI and Anthropic to withhold their models from the U.K.'s AI Security Institute until U.S. authorities cleared them, according to Bloomberg. That policy led Anthropic to stop British officials from reviewing its Claude Mythos 5.1 model, according to LBC, and Anthropic did not provide an explanation for withholding the model, according to Bloomberg.

"I'm very confident that, as a consequence of the conversations we've had, we will be able to test the latest generation of models," Narayan told Bloomberg. He also acknowledged the limits of the arrangement: "When you're reliant on someone else's technology, you have to be clear-eyed about the permanence of any outcome."

Founded in 2023, Britain's AI Security Institute ranks as the most generously resourced government body dedicated to assessing cutting-edge AI models, including those from OpenAI and Anthropic, according to Bloomberg. It is central to the U.K.'s broader ambition of staying relevant in AI development while carving out a political identity separate from the European Union, which has pursued more aggressive regulation of the technology.

In a speech on Monday at Imperial College London, Narayan also announced new measures to strengthen the U.K.'s cyber defenses using frontier AI. Two government pilots had used AI to identify more than 1,000 potential vulnerabilities in government systems and had been working to reduce the time taken to address them, the government said. Narayan said he would work with other ministries to find weaknesses in government systems before attackers could exploit them.

Narayan also warned that as AI systems become more capable of acting without human oversight, the risks of failures or misuse could grow. "If you cannot adequately control the risks, do not release the system," he said in a statement.

The access dispute is part of a broader tension between the U.K. and the U.S. over AI governance. Prime Minister Andy Burnham has cast the U.K. as an "honest broker" in shaping global AI standards, and has said AI will be a focus of the country's G20 presidency, according to LBC. President Donald Trump has said he would resist efforts to constrain what he has called "super intelligence," according to LBC.