As the UK pursues its ambitions to become an AI powerhouse it must contend with the mercurial nature of the second Trump administration.

If the UK government wanted a lesson about the fragility of plans to make the country an AI powerhouse, it might refer to the Trump administration’s decision to ban Anthropic from the federal supply chain. In February, US Defense Secretary Pete Hegseth shared that his department was designating Anthropic – an American AI company – a ‘supply chain risk’. Meanwhile, President Donald Trump ordered government departments to stop using Anthropic’s tools, such as the chatbot Claude, that are embedded in federal IT systems. ‘We don’t need it, we don’t want it, and will not do business with them again!’ wrote the President on social media.

In an open letter on Anthropic’s website, Chief Executive Officer Dario Amodei set out how its $200m contract with the Trump administration had unravelled. The company said it had refused to buckle under pressure to allow the Defense Department – now rebranded the Department of War – to use the technology in two specific cases that were ‘simply outside the bounds of what today’s technology can safely and reliably do,’ according to Amodei.

First, Amodei argued that the legal framework governing AI lagged behind current privacy laws. That meant it was possible for the government to use Anthropic’s technology to assemble ‘scattered, individually innocuous data into a comprehensive picture of any person’s life – automatically and at massive scale’. Second, he objected to an alleged demand by the Defense Department to allow it to use AI systems to develop and deploy weapons that could be used without human control. ‘Today, frontier AI systems are simply not reliable enough to power fully autonomous weapons,’ Amodei said. ‘We will not knowingly provide a product that puts America’s warfighters and civilians at risk.’

Emil Michael, Chief Technology Officer for the US Defense Department, has said that the uses of AI that concern Anthropic – mass domestic surveillance and fully autonomous weapons – are already barred by the law and by Pentagon policies. He added that the Pentagon’s reluctance to adopt contractual language proposed by Anthropic might relate to how the government may wish to respond to future developments concerning China.

Strong arm tactics

Following close analysis of Hegseth’s position in cutting Anthropic off from his department’s supply chain, some legal scholars have argued that the ban may not survive the rigours of a legal challenge, partly because it rests on two contradictory lines of attack. One says that Claude is so central to military operations that no restrictions on use are tolerable, while the other says that the AI tool is such a huge supply chain risk that the business must be purged from all government departments and even the cloud infrastructure it needs to operate those services.

‘What Hegseth is actually describing is not a supply chain risk determination but something closer to the beginning of a partial nationalisation of the AI industry,’ said Michael Endrias of Howard University School of Law and Alan Rozenshtein of the University of Minnesota. ‘Seize the technology and, if you can’t, destroy the company to ensure that no future AI developer dares negotiate terms the Pentagon dislikes.’

In March, Anthropic brought a lawsuit against the Defense Department for claiming that the company is a ‘supply chain risk’. In response, White House spokeswoman Liz Huston said that ‘our military will obey the United States Constitution – not any woke AI company’s terms of service’.

The legality of this case seems unimportant from a UK perspective, but the strategy that the US administration has employed should be a major cause for alarm. The UK government is attempting to position itself as a gateway destination for technology companies looking to bridge the AI divide between the US and Europe. It has taken several measures to smooth the way to this ambition, including designating data centres as critical national infrastructure, beefing up cybersecurity protection and trying to create data regulations that comply with strict EU requirements while appeasing US demands for looser, more favourable rules for its global technology companies.

Those measures didn’t go far enough to prevent the US government from suspending the flagship Technology Prosperity Deal with the UK, which was announced in September but then put on hold in December. President Trump was reportedly frustrated with the UK’s refusal to drop a two per cent sales levy on digital sales – affecting companies such as Google – and by a perceived lack of progress on agricultural trade.

The attempt by the US to fuse together negotiations on two separate trade areas may simply indicate that on the technology front, at least, the UK has very little to offer its larger counterpart – or it may represent a more aggressive attempt to force concessions in multiple sectors at once. So, while talks reportedly recommenced in February on how the two countries will collaborate on civil nuclear technologies, any discussions to move forward on AI and quantum computing remain on ice.

Stricter rules, higher risks

In fact, if the UK has been pursuing an appeasement agenda on technology regulation, that strategy may now be going into reverse. Enforcement action and regulations in the UK are moving towards stricter controls on US technology companies.

Early in 2026, for example, both the Information Commissioner’s Office (ICO) and Ofcom launched separate investigations into X. This was in relation to the AI chatbot Grok’s potential to produce harmful sexualised images and video content, following concerns about the chatbot’s ability to ‘undress’ images of real people. Grok is available to X users. While the ICO’s investigation, which is also looking into Grok’s provider xAI, covers the processing of personal data by the chatbot, Ofcom are scrutinising how this content has been shared on such a large scale on the platform. A negative outcome to either investigation could result in huge fines for the company – and a potential political reaction from the US administration.

In a statement issued in mid-January, X said it had implemented global measures to prevent Grok from allowing the editing of real people in revealing clothing. The company said it continued to ‘have zero tolerance for any forms of child sexual exploitation, nonconsensual nudity, and unwanted sexual content’ and would take action to remove such material.

Beyond the regulatory investigations, there has been significant legislative activity in this area in early 2026, with the UK government announcing that it will create new criminal offences relating to creating or requesting intimate, non-consensual images generated by AI or by other means. The UK Prime Minister Sir Keir Starmer announced in February, meanwhile, that his government would amend the law to close a legal loophole and bring all chatbots under the jurisdiction of the Online Safety Act. And in March, the government will launch a consultation on the issue of children’s wellbeing online, which will examine the possibility of measures such as minimum age limits for social media use.

Given this activity and its impact on tech companies, a clash with the US administration becomes increasingly likely. Back in early 2025, a White House statement outlined how the US government aimed to defend ‘American companies and innovators from overseas extortion and unfair fines and penalties’. In the case of any tax, regulation, act, policy or practice that negatively impacts such businesses, it said, the US government would retaliate by either imposing tariffs or taking any other action if it felt such intervention was needed to redress the balance.

That threat is best considered in light of Hegseth’s decision on Anthropic and President Trump’s threat to scrap trade with Spain over its government’s refusal to help US military strikes against Iran. ‘We’re going to cut off all trade with Spain. We don’t want anything to do with Spain,’ said President Trump at a White House press conference in March. The move looks like a consistent strategy to turn policy disagreements into commercial and economic threats. Again, such a move may not be legal in the long term, but the short-term impact could be severe and disruptive for trade between the two countries.

The UK isn’t alone in its over-dependency on US companies, with the EU recently backing a project to build sovereign cloud services on the continent to reduce its exposure to both American and Chinese companies.

Meanwhile in the UK, in December alone US tech company Palantir was reportedly awarded a £240m deal with the country’s Ministry of Defence without competition because no other supplier could offer comparable services and because of a need for compatibility with existing systems.

As Global Insight has previously reported, campaigners are unhappy about both Palantir’s involvement in developing software for the controversial US Immigration and Customs Enforcement agency, and with the UK government signing a health service contract to integrate patient data with the company before all the terms had been agreed. In response to these concerns, NHS England has highlighted that ‘the supplier will not control the data in the platform [in question], nor will they be permitted to access, use or share it for their own purposes’.

There are no European companies that have both the technology and the experience to match the largest US businesses in this area. So, without the Technology Prosperity Deal, the UK may struggle to build competitive technologies of its own. What the Anthropic episode shows is that both the UK and the CEOs of America’s most critical technology companies are vulnerable to maximalist demands by the Trump administration. No wonder most owners of the leading US technology companies have sought to deepen relationships with the incumbent government. And there’s little doubt who they’ll side with if the UK falls out of favour.

Arthur Piper is a freelance journalist. He can be contacted at arthur@sdw.co.uk