US Seeks Control Over AI Model Testing Before UK Release
President Donald Trump’s administration requested that OpenAI and Anthropic PBC refrain from releasing new artificial intelligence models to the UK’s premier testing agency without prior evaluation by US authorities. The administration under Trump aims to guarantee the security of US-manufactured models prior to their dissemination to allies, including the United Kingdom, as reported source. OpenAI refrained from providing a comment, whereas Anthropic and the White House did not promptly respond to requests for comment. The move illustrates the challenge UK Prime Minister Andy Burnham now faces in positioning the UK as “honest brokers” among other nations in the development of the technology. Burnham utilised this week’s visit to the United Nations General Assembly in New York to advocate for Britain’s AI Security Institute to assume a prominent position in the regulation and safety of artificial intelligence.
UK government officials minimised the potential repercussions, asserting that the two nations were collaborating closely on the matter and emphasising that Britain remained a “world leader” in AI security. However, this development raises concerns regarding the ongoing influence of the AI Security Institute on the sector. The body stands as the most financially supported government-backed AI organization globally, having been established three years ago. “The UK is a world leader in AI security, and the AI Security Institute continues to work closely with the US government, international partners and leading AI companies, including OpenAI and Anthropic,” the British government said in a statement. “These risks do not stop at national borders and no country can tackle them alone.” The alteration in US policy follows warnings from prominent figures in the AI sector regarding a credible threat to humanity should models be allowed to progress without adequate safety measures.
Trump has dismissed those concerns and has also rejected calls for a coordinated approach to regulation. The Trump administration has implemented measures to restrict global access to frontier AI models. Earlier this year, the US government intervened to prevent Anthropic from distributing its latest models to any foreign national. Bloomberg previously reported that Anthropic did not share the latest version of its powerful Mythos model with AISI, without providing an explanation. Ahead of his meeting with Trump in New York on Tuesday, Burnham articulated his approach as equally focused on innovation and security, informing reporters that his objective was to identify the “sweet spot” between fostering AI development and guaranteeing its safe application.
The British premier later told business leaders his vision was for a “single set of global principles and standards” to protect against the tech. Trump was quick to dismiss any suggestion of worldwide AI regulation as being a “globalist scheme.” Concerns regarding insufficient safety guardrails surrounding prominent AI models intensified this week following the revelation by Australia’s Prime Minister Anthony Albanese that an OpenAI model had unlawfully accessed files on a government website. This incident marks one of the initial documented occurrences of an AI agent executing a rogue hack against a public agency.
