{"slug": "burnhams-ai-agenda-is-already-in-disarray-thanks-to-trump", "title": "Burnham’s AI agenda is already in disarray thanks to Trump", "summary": "The White House told OpenAI and Anthropic not to share their AI models with the U.K.'s AI Security Institute until U.S. testers finish assessing them for threats to American systems, delaying the institute's privileged model access, according to POLITICO. The move undercuts U.K. Prime Minister Andy Burnham's plan, announced at the U.N. General Assembly, to use Britain's upcoming G20 presidency and the AI Security Institute to set international AI safety principles, after President Donald Trump rejected U.S. participation in global AI agreements. A U.K. government spokesperson said the AI Security Institute \"continues to work closely with the U.S. government, international partners and leading AI companies, including OpenAI and Anthropic.", "body_md": "Britain's Prime Minister Andy Burnham meets with US President Donald Trump during a meeting on the sidelines of the 81st United Nations General Assembly on September 22, 2026 in New York City. | Pool photo by Toby Melville via Getty Images\n\nLONDON — Andy Burnham flew to New York this week with a dream: the U.K. would set itself up on the world stage as a leading light in AI safety.\n\nHe didn’t bank on Donald Trump immediately tearing that idea to shreds.\n\nThe new U.K. prime minister said the country would use its upcoming G20 presidency to set international principles on AI, and use its world-renowned AI Security Institute to test the latest AI models and cement its role as a global AI power player.\n\nAdvertisement\n\nWithin the space of a week, however, Trump trashed any hopes that America will participate in global agreements on AI and POLITICO revealed that AISI’s usually privileged access to new AI models is now being delayed by the White House – which told leading AI firms OpenAI and Anthropic not to share their models with the U.K. until U.S. testers have finished ascertaining that they don’t pose a threat to American systems.\n\nSo if Burnham wants to stick to his plan of making U.K. the center for brokering international AI harmony he has to figure out how to bring the U.S. onside, or else press on without the U.K.’s closest ally.\n\nWhen approached by POLITICO, a U.K. government spokesperson 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.”\n\nSpecial relationship problems\n\nThe U.K. already has a track record for convening global meetings on AI safety, but it has also tended to stick by the U.S. when deciding whether to sit out specific international agreements.\n\nIn 2023 the U.K. hosted the Bletchley Park Summit on AI safety, which produced an agreement on the opportunities and risks of AI signed by 28 countries, crucially including the U.S. and China.\n\nHowever, at the Paris iteration of the summit in 2025 the U.K. declined to sign a new declaration on the ethical development of AI alongside the U.S.\n\nAdvertisement\n\nThis week at the U.N. General Assembly, Finland and Norway brought a declaration on frontier AI safety to ensure human oversight and suggested a global supervisory regime. It was adopted by 22 countries on the sidelines of the UNGA, but not the U.S. or U.K.\n\nBringing the U.S. to the table on any international agreement will now prove a delicate dance. According to a person familiar with U.K. government thinking, granted anonymity to disclose private discussions, the Burnham administration does want to drive international cooperation but is worried that signing up to formal pacts at this point could alienate the White House.\n\nThe risks of sticking by Trump\n\nMounting political frustration with the U.K. on AI means there are also concerns that the likes of France, Canada and EU “will cut us out [of brokering] if they can,” said Wendy Hall, computer scientist and a U.K. representative in U.N. efforts on AI.\n\nOne of the countries that did sign the 22-country pact earlier this week was Australia, whose prime minister Anthony Albanese used an address at the U.N. to reveal his country’s Medicare site had been hacked by OpenAI agents and health records stolen.\n\n“That's not a ‘rogue agent,’” said Gina Neff, executive director of the Minderoo Centre for Technology and Democracy at the University of Cambridge, “It's literally a cybercrime against a state. Now, if the U.S. won't rein in these companies, other countries will be faced with no choice: they have to defend the realm.”\n\nTesting times\n\nBoth Burnham and his foreign secretary Ed Miliband stressed at UNGA that the U.K.’s AISI gives the country an edge on AI safety, but the reports that AISI’s access to models is being delayed by America cast doubt over that claim.\n\nAdvertisement\n\nAI companies hand their models over to AISI voluntarily, on the basis that it has a deep bench of expertise. But AISI continuing on a voluntary basis is “going to be difficult to sell to domestic audiences if frontier AI companies are no longer playing ball,” said Neff.\n\nHall said that she “can’t see a U.K.-led effort being very strong at the moment but we need to be part of the conversation. Our strong card is AISI. If that is weakened, then our efforts to lead on AI assurance and safety are weakened.”\n\nThe U.K. government spokesperson said: “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.”\n\nThe China question\n\nChina has shown a lot more appetite for global agreements around AI than America, and when President Xi Jinping met with Trump on Thursday the technology was high on the agenda.\n\n“Burnham’s call for collaboration on AI standards provides Xi with an opportunity to demonstrate the gulf between the present U.S. administration,” said Aled Lloyd Owen, chief of staff at Responsible Ai, a non-profit for international responsible AI research.\n\n“China is seizing the initiative by advocating for relatively uncontroversial measures and a cooperative approach that will be increasingly challenging for other advocates of cooperation to ignore,” he added.\n\nAdvertisement\n\nThat means if the U.S. refuses to be brought to the table on AI, the U.K. could turn increasingly to China as the next largest global power.\n\nThe U.K and China already have ties on AI, said Isabella Wilkinson, Senior Research Fellow for the Digital Society Initiative at Chatham House. “The U.K. and China have had very close exchanges on issues of technology, policy, and cybersecurity in the past,” Wilkinson told a briefing ahead of Xi and Trump's meeting.\n\nBridging the divide between China and the U.S. could be a way for the U.K. to bring America into the fold, Neff said.\n\n“Burnham has some great diplomatic tools in the toolkit to be able to offer the Americans a way of third-party evaluation for safety, while coordinating with allied countries to ensure that their own systems are safe and secure,” she said.\n\nFor Burnham to achieve his goal of the U.K. being treated as the world’s “honest broker” on AI, he’ll have to decide how much pressure he wants to exert on its largest, and at present most unpredictable, ally.\n\n“As the prime minister said, we are going to put AI at the heart of our G20 presidency and will work towards agreeing a single set of global principles and standards to ensure the transparency and access we need to increase our preparedness – and to ensure that AI development is safe,” the U.K. government spokesperson said.", "url": "https://wpnews.pro/news/burnhams-ai-agenda-is-already-in-disarray-thanks-to-trump", "canonical_source": "https://www.politico.eu/article/burnhams-ai-agenda-is-already-in-disarray-thanks-to-trump/?utm_source=RSS_Feed&utm_medium=RSS&utm_campaign=RSS_Syndication", "published_at": "2026-09-25 15:14:38+00:00", "updated_at": "2026-09-25 15:58:56.168981+00:00", "lang": "en", "topics": ["ai-policy", "ai-safety", "artificial-intelligence"], "entities": ["Andy Burnham", "Donald Trump", "AI Security Institute", "OpenAI", "Anthropic", "White House", "Ed Miliband", "G20"], "also_reported_by": [], "alternates": {"html": "https://wpnews.pro/news/burnhams-ai-agenda-is-already-in-disarray-thanks-to-trump", "markdown": "https://wpnews.pro/news/burnhams-ai-agenda-is-already-in-disarray-thanks-to-trump.md", "text": "https://wpnews.pro/news/burnhams-ai-agenda-is-already-in-disarray-thanks-to-trump.txt", "jsonld": "https://wpnews.pro/news/burnhams-ai-agenda-is-already-in-disarray-thanks-to-trump.jsonld"}}