On 12 June 2026, the Trump administration ordered Anthropic to cut off all foreign nationals from Fable 5 and Mythos 5 — its two most powerful models — overnight.
Unable to verify citizenship in real time, Anthropic simply switched both off globally.
The stated reason? A narrow jailbreak found by Amazon researchers. TechCrunch reported the real driver was "personality differences" between Anthropic and the administration. Anthropic warned the move would "essentially halt all new model deployments" industry-wide.
Economic Nationalism Dressed in Security Language?
Tom Tugendhat was blunt: "Sovereignty is more about code than cannons." A British official added: "British researchers, companies and hospitals were using this. Not any more."
The contradiction is hard to ignore. The US relaxed chip export controls (~$34B/month) to fuel its AI boom, while cutting off close allies from its best models. Chatham House warned in April that AI export controls are "not the best bargaining chip": enforcement is near-impossible and allies are being treated as afterthoughts.
A Smoot-Hawley Moment?
The 1930 Smoot-Hawley Act triggered retaliation and deepened the Great Depression. That dynamic is already repeating:
China's 2025 chip restrictions sent Volvo and Volkswagen into crisis. France, Germany and the UK are now accelerating sovereign AI investment, fragmenting the global ecosystem and slowly eroding US commercial dominance.
The UK's Uncomfortable Position
Kier Starmer spent the weekend lobbying Washington for a British exception. His AI adviser, Jade Leung, is reported to be pressing for Britain to be included on any future "approved countries" list, whilst his AI minister has called for "sovereign AI capability."
This puts the UK in a difficult position. On the one hand, publicly lobbying an ally for access to commercial technology that it is increasingly reliant upon — whilst acknowledging it controls none of the supply chain, has no home-grown frontier model, and seriously lacks the compute required to train one, even if it had one.
The government has agreed a £1.1bn AI hardware plan, which is welcome but seems like a drop in the ocean against the deep pockets of Altman, Amodei, Musk, Pichai and Huang.
The Question Every CIO Should Be Asking
This episode exposed a dependency most organisations hadn't fully reckoned with: your AI capability can be switched off by a foreign government without warning.
That's not a technology risk. It's a geopolitical one — and it belongs in your risk register.
Should one country be able to pull the plug on your digital workforce? I'd be interested in your thoughts — connect on LinkedIn or get in touch.
Further reading: TechCrunch — The US government's Anthropic models ban was never about an AI jailbreak · The Telegraph — Britain cut off from advanced AI after Trump ban