US asks Anthropic to block global access to top AI models: Why it matters

Artificial intelligence has become the latest issue to drive a wedge between the United States and its allies after US President Donald Trump ordered tech giant Anthropic to cut off foreign access to its powerful Mythos 5 and Claude Fable 5 AI models, citing national security concerns.

The US issued the unprecedented order for all foreign nationals in and outside the US last week, promoting Anthropic to take the two AI models completely offline to ensure compliance.

Recommended Stories

list of 4 items list 1 of 4 Luis Romo strikes as Mexico become first team to reach World Cup knockouts

list 2 of 4 Behind the noise of an ‘Iran deal’, Palestine continues to burn

list 3 of 4 Burnham by-election victory raises stakes for Starmer

list 4 of 4 Cape Verde goalkeeper Vozinha happy his mother will attend World Cup

end of list Anthropic had granted 200 institutions across 15 countries access to their frontier model, Claude Mythos Preview, to test for vulnerabilities.

The two public versions of the model, Mythos 5 and Fable 5, were due to be released in early June.

Anthropic said the US government did not provide a reason for the order, but that it was its “understanding” that the Trump administration believed it had become aware of a method of “jailbreaking” Fable 5.

The Trump administration’s ban immediately sent shockwaves across Europe, which is heavily dependent on US-developed AI.

French President Emmanuel Macron told a meeting of the Group of Seven (G7) nations this week that while the Trump administration’s order was a “wake-up call” about the dangers of AI, the limits were a “bad thing”.

“The reaction is in some regards strictly nationalist,” Macron said on Wednesday.

While the US has targeted adversaries like China and Russia with numerous tech restrictions, the Trump administration’s Anthropic order applies equally to allied countries that have intelligence-sharing and mutual defence pacts with Washington.

The decision was a first for the AI industry, but it comes on the heels of other transactional policy moves by the Trump administration.

Over the past 18 months, Trump has launched a global trade war, and threatened to annex Greenland, an autonomous Danish territory, and withdraw from the 77-year-old North Atlantic Treaty (NATO) alliance.

He also threatened to stop supplying weapons to Ukraine unless European allies helped reopen the Strait of Hormuz, which was in effect shuttered by Iran after the US and Israel launched their war on the country on February 28.

French President Emmanuel Macron and Anthropic CEO Dario Amodei attend a working lunch with G7 leaders, G7 outreach partners and global tech CEOs on innovation and AI during the G7 Summit in Evian-les-Bains, France, on June 17, 2026 [Evelyn Hockstein/Reuters] US allies are now waking up to the realisation that they are “far too vulnerable now to the US techno-industrial complex,” Dex Hunter-Torricke, president of the Center for Tomorrow, told Al Jazeera.

Macron also stressed the need for countries to work together on addressing AI issues, warning against the danger posed by “non-cooperation between democracies”.

Macron’s remarks were echoed by Thomas Regnier, European Commission spokesperson for tech sovereignty, who told Al Jazeera that addressing security concerns was a “shared challenge, not one confined to a single jurisdiction or country”.

The solution should also not be “discriminatory against partners,” Regnier said.

At the closed-door meeting, the G7 countries – the leaders of Canada, France, Germany, Italy, Japan, the United Kingdom, and the US – discussed a potential “trusted partner” scheme for access to the most advanced AI models, though few details have been disclosed.

The US briefly introduced a similar tiered model for the semiconductor chips that power AI in early 2025 during the last weeks of US President Joe Biden’s presidency .

Known as the “small yard, high fence” model in public policy circles, the scheme was intended to keep the most advanced US tech away from the likes of China and Russia, but it also created some “disquiet” among allies, said David Smith, an expert in US politics and foreign policy at the University of Sydney’s United States Studies Centre.

The scheme did not restrict allied access to semiconductors, but set out to restrict what they could do with those chips, including in a commercial setting or in their trade dealings with China.

“This was controversial because of the fact that in the past, when the US has restricted the export of certain technologies, they have always been things with direct military applications,” Smith told Al Jazeera.

“Restricting access to advanced chips was very different.”

An NVIDIA logo and a computer motherboard appear in this illustration taken on August 25, 2025 [Dado Ruvic/Reuters] The scheme was scrapped in May 2025 by the Trump administration – which later cleared the powerful Nvidia H200 chip for sale to a limited number of Chinese firms.

The Anthropic ban is, meanwhil…