Artificial Intelligence & Machine Learning , Geo-Specific , Next-Generation Technologies & Secure Development

The U.S. government’s decision to cut foreign nationals’ access to Anthropic’s most powerful artificial intelligence models has sparked a massive increase in calls for Europe to reduce its reliance on American technology.

See Also: How AI Increases the Risk of Enterprise Data Exposure

The abrupt introduction of export controls for the Mythos and Fable models prompted Anthropic to immediately disable access for everyone. The government-imposed restrictions apply to foreign nationals in the United States as well as abroad, with even Anthropic employees being affected. Mythos has strong capabilities for finding and exploiting software vulnerabilities, and Fable was a new publicly-accessible version that was supposed to have strong cybersecurity guardrails (see: US Pulls the Plug on Anthropic’s Top AI Models).

Anthropic said that someone had told the government that Fable’s guardrails could be bypassed - Reuters and others reported that Amazon CEO Andy Jassy was among those who raised such concerns with the Trump administration. Anthropic said the reported "jailbreaking" technique could only be used to find simple vulnerabilities that even more primitive models could spot, and insisted that Fable’s safeguards were "substantially more effective than those of any previously deployed model."

Either way, the world lost access to powerful AI models, and the reaction in the European Union and the United Kingdom has been explosive.

"This development is a further illustration of why Europe needs to strengthen its technological sovereignty, and it underlines the relevance of the cybersecurity and AI legislation already in place at the EU level, including the AI Act, the Cyber Resilience Act and the NIS2 Directive - as tools to manage exactly this kind of risk on our own terms," said Thomas Regnier, the European Commission’s spokesperson on tech sovereignty issues, in an emailed statement.

Although they do all cover cybersecurity risks to varying degrees, none of the laws Regnier mentioned have much to say about reducing dependence on foreign technology. The commission did earlier this month unveil a long-delayed "tech sovereignty package" of legal proposals, including a law called the Cloud and AI Development Act that would effectively shut U.S. Big Tech out of public cloud contracts involving the processing of sensitive information (see: EU Prepares Path for Shutting Out US Cloud Providers).

Tech sovereignty has become a live topic in Europe since the advent of the second Trump administration, due to repeated American threats to annex Denmark-owned Greenland, and fears that Washington could flip a "kill switch" on its widely-used technology if it so wished. Some say the Anthropic action is an example of just that.

Benjamin Haddad, the French minister delegate for European affairs, said in a Saturday post on social media network X that the Trump administration's Mythos move accelerated the "geopolitical battle over AI" and demonstrated how "the United States now openly embraces AI as a strategic instrument tied to national security, a weapon of domination not to be placed in all hands." He added that the episode showed how critics of the EU’s tech sovereignty push - which France is embracing more than most - "must face reality."

Jordan Bardella, leader of the far-right National Rally party and possibly the next French president, also weighed in. "This sudden decision comes to remind us that AI is already a major issue of national sovereignty. Nations that do not quickly develop their own models will always depend more and more on the choices of other powers," he said in an X post, adding that France should "accelerate its support" for Paris-based Mistral AI, which is the EU's only competitor to Anthropic and OpenAI.

Similar sentiments were expressed across the English Channel - although, unlike France, the U.K. has no significant AI player to promote.

"This week the most advanced AI in the world was cut off from everyone in Britain, not by us, but by a decision taken in another country," said Kanishka Narayan, the U.K.’s minister for AI and online safety, in a video on X. "We treat every other threat to our sovereignty with deadly seriousness, but we haven’t learned to treat this one in the same way."

Narayan said the future of British AI was "the most significant question for Britain’s national security." However, he added: "I’m not going to pretend there’s a simple switch we can pull. There isn’t. Britain needs more AI capability. This is the central political question of our time and our first duty is to see it clearly before someone else decides the answer for us."

The U.K.’s AI Safety Institute was one of the first non-U.S. organizations to gain access to Mythos after Anthropic announced the model’s existence in early April. The institute has reported that Mythos did indeed represent a significant leap forward in AI cybersecurity capabilities, as does OpenAI’s GPT-5.5-Cyber model.

It took nearly two months for Anthropic to agree to give EU organizations similar access - the reasons for its previous reluctance have never been explained - and ENISA, the EU’s cybersecurity agency, told ISMG two weeks ago that negotiations were ongoing as to the terms of potential access (see: Europe Edges Closer to Claude Mythos Access).

A spokesperson for ENISA said on Monday that Anthropic had informed the agency about the export directive on Friday, and it was now discussing the matter with its partners at the Commission and in EU countries.

The circumstances surrounding the U.S. government’s decision remain murky. Anthropic already had a fractious relationship with the administration, apparently stemming from its unwillingness to let the government use Mythos for fully autonomous weapons systems or mass domestic surveillance. Former U.S. AI czar David Sacks said in a Saturday X post that there was no link between the two matters. Sacks said the government issued the export controls because Anthropic "hasn’t wanted to cooperate with a reasonable safety request (i.e. fixing the jailbreak issue)."

President Trump repeatedly said this month that he would like the government to get stakes in the country’s big AI companies. Both Anthropic and OpenAI are preparing to float on the public markets following Friday’s bumper IPO for SpaceX, the parent company of xAI and therefore also a significant player in the sector.

The Financial Times reported earlier this month that the U.S. National Scurity Agency was using its Mythos access to launch offensive cyber operations.