Trump to reject UK plea over Anthropic ban as AI ‘kill switch’ fears grow
The White House is poised to reject Keir Starmer’s attempt to restore British access to Anthropic’s most advanced AI models, in a move that has exposed how dependent the UK has become on a handful of American Big Techs.
Downing Street has spent recent days lobbying Washington for an exemption after the Trump administration blocked foreign access to Anthropic’s Fable 5 and Mythos 5 systems, two of the company’s most powerful AI models – but hopes of securing a carve-out are fading fast.
According to the Telegraph, one source familiar with the matter said there was “zero chance” of Britain receiving special treatment, while administration officials have indicated that any eventual rollback is more likely to apply globally than be granted specifically to allies such as the UK.
The dispute has quickly become about more than a single AI model, highlighting a growing vulnerability at the heart of Britain’s tech sector, where critical infrastructure increasingly sits in the hands of overseas providers who can switch off access with little warning.
The restrictions were introduced after US officials raised concerns that Anthropic’s recently released Fable 5 model could potentially be “jailbroken”, allowing users to unlock the more powerful cyber-security capabilities of Mythos 5.
But Anthropic has strongly disputed that assessment, arguing instead that the evidence presented by Washington relates only to a narrow vulnerability.
The company has warned that applying the same standard across the industry could effectively halt the release of frontier AI models altogether.
The disagreement triggered emergency talks in Washington on Monday between Anthropic executives and officials from the US Commerce Department. And according to people briefed on the discussions, the meeting ended without the restrictions being lifted.
US Commerce Secretary Howard Lutnick reportedly joined parts of the talks remotely from the G7 summit in France, while Anthropic representatives sought to persuade officials that the security concerns had been overstated – yet the administration remained unconvinced.
A wake-up call for the UK
The episode has intensified debate around Britain’s ability to build its own AI capabilities rather than relying on American providers.
Kanishka Narayan, the UK’s AI minister, described the restrictions as evidence of why Britain needs greater “sovereign AI capability“, while business secretary Peter Kyle pointed to the need to accelerate domestic tech innovation.
“This week, the most advanced AI in the world was cut off for everyone in Britain,” Narayan said.
Starmer is expected to raise the issue directly with Trump during meetings at the G7 summit, although officials privately acknowledge there is little appetite in Washington to create country-specific exemptions.
It has been reported that the original intention was not to target Britain or Europe specifically, but rather to force Anthropic to withdraw the models entirely while concerns over their capabilities were assessed.
But dozens of cyber-security experts, including staff from Nvidia, Zoom and Mercedes-Benz, have signed an open letter calling for the restrictions to be lifted, warning that removing advanced tools from legitimate security researchers could make Western networks less secure rather than more secure.
Anthropic has long occupied a complicated position in Washington, where it has repeatedly warned about the risks posed by advanced AI systems while simultaneously developing some of the world’s most powerful models.
The latest row comes after the Pentagon reportedly severed ties with the company earlier this year over disagreements surrounding the use of its tech.
Anthropic first unveiled the Mythos 5 model in April, granting access only to governments and a limited group of organisations due to its unusually advanced ability to identify cyber vulnerabilities.
Last week, it released a constrained public version known as Fable 5 before US officials intervened.