Austria Urges Anthropic to Move to EU to Avoid US Controls

Agentic AI , Artificial Intelligence & Machine Learning , Governance & Risk Management Austria Urges Anthropic to Move to EU to Avoid US Controls Mythos and Fable Export Controls Deprive EU of 'Cutting-Edge Innovation,' Security David Meyer • June 29, 2026     Credit Eligible Get Permission In response to US export controls on its AI models, the Austrian government wants Europe to tempt Anthropic into moving across the Atlantic. (Image: Shutterstock) Stung by the Trump administration's export controls on Anthropic's most powerful cyber-capable artificial intelligence models, Mythos and Fable, the Austrian government wants Europe to tempt Anthropic into moving across the Atlantic. See Also: Edge Transformation: Top 5 SASE Predictions and Trends "Let us jointly explore the strategic establishment and participation of Anthropic within the European Union - with legal certainty, market access, capital and a set of values that suits this company," Alexander Pröll, the Austrian state secretary for digitalization, wrote in a letter sent Saturday afternoon to Henna Virkkunen, the European Commission's vice president for tech sovereignty. On June 12, the U.S. government invoked export controls to bar Anthropic's most advanced AI models, Fable 5 and Mythos 5, from being accessed by foreign nationals worldwide, citing national security concerns. The move forced the company to temporarily suspend the models globally because it couldn't selectively restrict access. "Overnight, the world's largest single market - our single market, comprising 450 million people - was cut off from a cutting-edge innovation. Not by our decision. But by that of a foreign government," Pröll wrote. "This decision was not to be dismissed as a footnote. It was a show of power. And it serves as a reminder of how vulnerable we in Europe are when the key technology of our time lies beyond our sphere of influence and access to it can be withdrawn at the stroke of a pen." Pröll said the EU has a short window to lure Anthropic from Silicon Valley and said, "If we want to act, we must act now." "The question is whether we Europeans are prepared to be the architects of our technological future, or whether we wish to remain mere administrators of decisions made elsewhere," he said. Vincenz Kriegs-Au, Pröll's spokesperson, confirmed to ISMG on Monday that the state secretary was talking about a wholesale move on Anthropic's part, not merely the opening of a European subsidiary. The company already has EMEA headquarters in Dublin, with additional EU offices in Paris, Munich and Milan. "Think big," he said. In a midday press briefing, European Commission spokesperson Thomas Regnier confirmed receipt of Pröll's letter and suggested it would be Anthropic's decision where to locate itself. Anthropic didn't immediately responded to a request for comment from ISMG. The European push for tech sovereignty has stepped up significantly since the second Trump administration threatened to annex Greenland, an autonomous territory of Denmark. The last 18 months have seen much discussion about the possibility of the U.S. president flipping a "kill switch" on the American tech services that Europe needs, and the Anthropic export controls have been widely perceived as an example of that power. Anthropic's Claude Mythos model, unveiled nearly three months ago, is particularly skilled at finding software vulnerabilities and even exploiting them. OpenAI's newer GPT-5.5-Cyber model has similar capabilities. Anthropic has only made Mythos available to a select group of organizations through an initiative called Project Glasswing, so they can use it to harden their cyber defenses. It took nearly two months for Anthropic to invite European organizations including the European Commission into that fold (see: Europe Edges Closer to Claude Mythos Access.) But before Europeans could actually get access to Mythos, the White House effectively pulled the plug on the initiative. In early June, Anthropic released a version of Mythos to the general public in the form of Fable, which supposedly had guardrails to prevent its use in cyberattacks. But someone - reportedly Amazon CEO Andy Jassy - complained to the administration that these guardrails were easily circumvented through jailbreak attacks, so the government suddenly instituted export controls on both Mythos and Fable (see: US Anthropic Export Controls Spark Sharp EU Reaction). On Friday, the Trump administration announced that it was partly rolling back the controls on Mythos, as Anthropic had worked to address the security risks and "committed to work with the U.S. government on protocols and standards and releases" for Mythos and Fable. Fable 5 remains unavailable to the public inside and outside the United States. On the same day, OpenAI said the government had asked it to limit the rollout of its new GPT-5.6 model, so it released it to a small group of users, which was first reviewed by the Trump administration. It doesn't appear that European organizations such as the European Commission are among those who may now gain access to Mythos, or who will be able to participate in the limited GPT-5.6 preview. Regnier said Monday that "the EU should have a say on how to best cooperate" on the security risks surrounding Mythos and hacker-friendly models like it. "We believe that we are a trusted partner, not a security risk," he said. In response to Pröll's letter, Regnier insisted that the European Commission had not been "left out" and "has done a great job when it comes to discussing bilaterally" with Anthropic and OpenAI. Europe's tech sovereignty drive is intended not only to reduce dependency on the United States and China, but also to stimulate local industrial development. The European Commission has proposed a tech sovereignty law that would effectively make European cloud providers the only option for certain high-risk public-sector use cases, and France has recently marshalled nearly $15 billion in funding for French and European tech companies (see: France and Germany Boost Digital Sovereignty Push). The continent does already have some sizeable AI companies, particularly France's Mistral, which is the continent's only real rival to Anthropic and OpenAI in the frontier model space - albeit a rival that remains substantially behind. Pröll insisted in his letter that trying to attract Anthropic is "not about undermining" these firms, which "deserve our full support." "A company of Anthropic's caliber on European soil would not displace our own champions. It would attract talent, retain capital, set standards and generate a pull that would benefit the entire European innovation landscape," Pröll wrote. "It is precisely this potential and the associated ecosystem that is at stake." U.S. officials including Vice President J.D. Vance have criticized Europe for being a heavy-handed regulator in the AI space. But Pröll claimed Anthropic would be "a good fit" for the continent because it shares the sentiment about prioritizing "safety over speed." "That is a profoundly European attitude," he wrote. "We should have the courage to say this out loud: This company would not be constrained in Europe. It would be set free."