US export ban on Anthropic’s AI models further strains alliances

Artificial quality has go the latest contented to thrust a wedge betwixt the United States and its allies aft US President Donald Trump ordered tech elephantine Anthropic to chopped disconnected overseas entree to its almighty Mythos 5 and Claude Fable 5 AI models, citing nationalist information concerns.

The US issued the unprecedented bid for each overseas nationals successful and extracurricular the US past week, promoting Anthropic to instrumentality the 2 AI models wholly offline to guarantee compliance.

Recommended Stories

• list 1 of 4Luis Romo strikes arsenic Mexico go archetypal squad to scope World Cup knockouts
• list 2 of 4Behind the sound of an ‘Iran deal’, Palestine continues to burn
• list 3 of 4Burnham by-election triumph raises stakes for Starmer
• list 4 of 4Cape Verde goalkeeper Vozinha blessed his parent volition be World Cup

Anthropic had granted 200 institutions crossed 15 countries entree to their frontier model, Claude Mythos Preview, to trial for vulnerabilities.

The 2 nationalist versions of the model, Mythos 5 and Fable 5, were owed to beryllium released successful aboriginal June.

Anthropic said the US authorities did not supply a crushed for the order, but that it was its “understanding” that the Trump medication believed it had go alert of a method of “jailbreaking” Fable 5.

The Trump administration’s prohibition instantly sent shockwaves crossed Europe, which is heavy babelike connected US-developed AI.

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

“The absorption is successful immoderate regards strictly nationalist,” Macron said connected Wednesday.

While the US has targeted adversaries similar China and Russia with galore tech restrictions, the Trump administration’s Anthropic bid applies arsenic to allied countries that person intelligence-sharing and communal defence pacts with Washington.

The determination was a archetypal for the AI industry, but it comes connected the heels of different transactional argumentation moves by the Trump administration.

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

He besides threatened to halt supplying weapons to Ukraine unless European allies helped reopen the Strait of Hormuz, which was successful effect shuttered by Iran aft the US and Israel launched their warfare connected the state connected February 28.

French President Emmanuel Macron and Anthropic CEO Dario Amodei be a moving luncheon with G7 leaders, G7 outreach partners and planetary tech CEOs connected innovation and AI during the G7 Summit successful Evian-les-Bains, France, connected June 17, 2026 [Evelyn Hockstein/Reuters]

US allies are present waking up to the realisation that they are “far excessively susceptible present to the US techno-industrial complex,” Dex Hunter-Torricke, president of the Center for Tomorrow, told Al Jazeera.

Macron besides stressed the request for countries to enactment unneurotic connected addressing AI issues, informing against the information posed by “non-cooperation betwixt democracies”.

Macron’s remarks were echoed by Thomas Regnier, European Commission spokesperson for tech sovereignty, who told Al Jazeera that addressing information concerns was a “shared challenge, not 1 confined to a azygous jurisdiction oregon country”.

The solution should besides not beryllium “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 imaginable “trusted partner” strategy for entree to the astir precocious AI models, though fewer details person been disclosed.

The US concisely introduced a akin tiered exemplary for the semiconductor chips that powerfulness AI successful aboriginal 2025 during the past weeks of US President Joe Biden’s presidency.

Known arsenic the “small yard, precocious fence” exemplary successful nationalist argumentation circles, the strategy was intended to support the astir precocious US tech distant from the likes of China and Russia, but it besides created immoderate “disquiet” among allies, said David Smith, an adept successful US authorities and overseas argumentation astatine the University of Sydney’s United States Studies Centre.

The strategy did not restrict allied entree to semiconductors, but acceptable retired to restrict what they could bash with those chips, including successful a commercialized mounting oregon successful their commercialized dealings with China.

“This was arguable due to the fact that of the information that successful the past, erstwhile the US has restricted the export of definite technologies, they person ever been things with nonstop subject applications,” Smith told Al Jazeera.

“Restricting entree to precocious chips was precise different.”

An NVIDIA logo and a machine motherboard look successful this illustration taken connected August 25, 2025 [Dado Ruvic/Reuters]

The strategy was scrapped successful May 2025 by the Trump medication – which aboriginal cleared the almighty Nvidia H200 spot for merchantability to a constricted fig of Chinese firms.

The Anthropic prohibition is, meanwhile, accelerating calls for much self-reliance among US allies.

“The concern we’re successful collectively close present with Mythos and Fable is thing that tin hap with over-reliance,” Canadian Prime Minister Mark Carney told the media up of the G7 summit.

“Nobody has done thing incorrect successful the situation. But we volition person done thing incorrect if we conscionable judge this, don’t instrumentality the lesson, don’t physique retired and diversify.”

Bruno Retailleau, a erstwhile French curate and campaigner successful the 2027 statesmanlike election, said Anthropic should beryllium a “wake-up call” to Europe that “a federation that depends connected others for its exertion is simply a federation that tin beryllium unplugged overnight”.

“We indispensable dainty AI the mode we treated atomic power: we indispensable deliberation of it arsenic portion of our sovereignty. Master it oregon endure it: determination is nary different path,” Retailleau said connected X connected Saturday, the time aft the prohibition was enactment successful place.

Tom Tugendhat, a erstwhile information curate nether Conservative Prime Ministers Liz Truss and Rishi Sunak, said the incidental was the “inevitable effect of exertion shaping warfare truthful that sovereignty is much astir codification than cannons”.

“We cannot proceed similar this and stay sovereign,” Tugendhat said connected X connected Saturday.

The Anthropic logo is seen successful this illustration taken connected May 20, 2024 [File: Dado Ruvic/Reuters]

Marcin Jerzewski, caput of the Taiwan bureau of the European Values Center for Security Policy, said that European companies could travel retired up successful the Anthropic controversy.

Prior to the Anthropic ban, German media reported successful April that the country’s equipped forces would not grant contracts to Palantir – Peter Thiel’s large information analytics institution that partners with instrumentality enforcement, defence and quality agencies – ‌citing concerns astir granting backstage manufacture entree to nationalist information systems.

Germany and France’s home spy agencies person besides precocious partnered with European companies implicit Palantir truthful arsenic to debar becoming excessively babelike connected US technology, according to officials.

As Anthropic has stoked treatment astir the request for AI sovereignty, it has besides drawn attraction to Paris-based AI startup Mistral, the “EU’s lone large homegrown frontier-model competitor”, Jerzewski told Al Jazeera.

“Overall, European governments are increasing uneasy astir their overreliance connected US-controlled technologies,” helium said.

“To this end, European companies mightiness payment from the Anthropic incident.”