Austria wants the EU to host Anthropic — and it exposes a bigger problem
Austria has formally asked the European Commission to explore hosting Anthropic — the maker of Claude — within the EU, after US export controls in June 2026 knocked out access to the company's two most advanced AI models globally. The move, triggered by the US Commerce Department restricting Anthropic's Fable 5 and Mythos 5 models on national-security grounds, signals how fragile European access to frontier AI has become. For British and American users and businesses that relied on those models, the disruption was immediate and unplanned.
What actually happened
On June 12, 2026, Washington imposed nationality-based export controls on Fable 5 and Mythos 5. Anthropic found that screening users by nationality was unworkable — it would have meant dismissing foreign staff — so it pulled both models offline entirely, for everyone. A partial fix arrived on June 26, when vetted US institutions regained access to Mythos, but Fable 5 remained restricted and access for non-US users, including the UK, did not resume. The episode confirmed what many had suspected: a single policy shift in Washington can sever access to critical AI infrastructure overnight, with no warning.
Austria's pitch — and why it's complicated
Austria's State Secretary for Digitalization Alexander Pröll sent an official letter to Virkkunen — EU Tech Commissioner Henna Virkkunen — arguing that Europe should offer Anthropic legal certainty, market access, capital, and a values alignment that Washington apparently can't guarantee. The idea is that anchoring Anthropic in the EU would insulate European operators from future US policy swings and act as a catalyst for local AI talent.
The pitch is politically tidy but practically thorny. Anthropic runs on Amazon Web Services infrastructure under a roughly $100 billion compute deal — moving jurisdiction doesn't move the data centers. EU membership also means full exposure to the AI Act, whose compliance costs for high-risk systems run to an estimated €29,000–€34,000 per unit annually. The UK faced the same access problem from a weaker position: post-Brexit, British regulators have no direct say over Anthropic's operations and lack the EU's bloc-level negotiating leverage.
The broader signal
Anthropic has been expanding its European footprint — it recently hired a senior Orange executive to lead European operations and is in talks with EU cybersecurity agency ENISA over the Mythos cyber model. But no formal response to Austria's proposal has emerged. The episode is less a workable relocation plan than a warning flare: Europe's dependence on US-controlled AI infrastructure is a structural vulnerability, and the "special relationship" — for the UK especially — is no substitute for regulatory leverage.
