According to Politico, citing a high-ranking American official and a person "familiar with the matter," the White House has reportedly asked Anthropic and OpenAI to temporarily halt sharing new artificial intelligence (AI) models with the British government. This request reportedly comes from the National Cybersecurity Center, a U.S. agency focused on protecting digital infrastructure. The two companies, which develop some of the most advanced AI systems in the world, had planned to test these models with the UK’s Artificial Intelligence Security Institute (AISI), an agency known for its strong resources and expertise in AI safety.
The decision to pause sharing has reportedly caused some unease among the companies. Anthropic’s most advanced model, called Mythos 5.1, remains exclusive to the United States for now. The U.S. administration says its goal is to ensure that its own systems are secure before allowing these models to be used by international partners. A high-ranking U.S. official told Politico, “These are American companies, and it has always been our policy with every new cutting-edge model launched.” The White House has not officially commented on the matter.
Anthropic has stated in a blog post that it is working with the U.S. government to gradually expand access to its models for both national and international partners. However, the company has not confirmed or denied the Politico report. Meanwhile, Henry de Zoete, director of the AISI, wrote to British parliamentarians about maintaining strong relationships with top AI firms. He confirmed that the AISI had access to GPT-6 Astra, a model from OpenAI, before its official release on September 3, but did not specify how long in advance.
British Prime Minister Andy Burnham has downplayed any concerns about U.S. distrust, though he emphasized that AI will be a key topic at the upcoming G20 summit, which the UK will host. The situation highlights the growing importance of AI in global diplomacy and security, as well as the delicate balance between innovation and national interests.
U.S. Presses Anthropic and OpenAI to Pause AI Model Sharing with UK
AI-rewritten from original reportingHow it works
aisecurityusaukanthropicopenai



