Austria has called on the EU to formally explore establishing Anthropic within the bloc, as the US moves to restrict non-American users from accessing the AI company’s most advanced models.
The move arrives at a moment of growing anxiety across the continent about what it means to be dependent on American AI infrastructure, particularly when the rules governing that access can change overnight. With Europe already pushing to reduce its reliance on US technology more broadly, Austria’s intervention signals that the AI sovereignty debate has moved well beyond policy papers.
Although such a move may not seem achievable, many argue that the more pressing question is whether Europe intends to shape its own technological future or remain subject to decisions made elsewhere.
Inside the Proposal
The proposal, set out in a letter from Austria’s State Secretary for Digitalization Alexander Proell to EU Technology Commissioner Henna Virkkunen, represents one of the most direct interventions yet by a member state on the question of Europe’s access to frontier AI.
In his letter, Proell called for a joint exploration of the strategic establishment and participation of Anthropic within the EU, pointing to legal certainty, market access, capital, and shared values as the foundations Europe could offer. The letter does not prescribe a specific mechanism, whether a European subsidiary, a licensing arrangement, or a deeper structural presence. Instead, it frames the question as one of strategic urgency rather than commercial convenience.
The proposal was triggered by a US directive that forced Anthropic to disable access to its most advanced models, Fable 5 and Mythos 5, for foreign nationals. The restriction landed without warning or recourse, cutting off European businesses, researchers, and public institutions that had built workflows around the company’s systems.
The scale of the disruption highlighted just how exposed Europe’s dependency had become. Access to two of the most capable AI models on the market was severed by a policy decision taken in Washington, one that European institutions had no mechanism to challenge, delay, or appeal.
The European Commission had itself proposed new legislation earlier this month aimed at strengthening domestic cloud, AI, and semiconductor industries while reducing reliance on US Big Tech. Austria’s letter can be read as a direct extension of that momentum, pushing the conversation from legislative intent to concrete industrial strategy.
A Fraught Backdrop
Austria’s proposal cannot be read in isolation from a period of significant tension in Anthropic’s relationship with the US government. The company had previously been labeled a supply chain risk following a dispute with the White House over how its models could be used and deployed, a confrontation that illustrated how quickly AI companies can become caught between commercial interests and national security imperatives.
The subsequent restriction of its Mythos model to US nationals only marked an escalation of that dynamic, with immediate consequences for European users who had no warning the restriction was coming. For enterprises and public bodies that had integrated Anthropic’s tools into critical workflows, the disruption was both operational and political.
That context reframes how the Austrian proposal should be understood. The EU has spent recent years building a digital sovereignty agenda, encouraging institutions and businesses to favor European technology providers and reduce their exposure to American platforms. That push has been driven by concerns over data residency, regulatory jurisdiction, and the long-term strategic risks of concentrating so much capability among a handful of US hyperscalers.
The difficulty is that European AI companies have not kept pace with the frontier. No EU-based model currently competes with Anthropic’s Claude, OpenAI’s GPT-4o, or Google’s Gemini at the highest capability levels, leaving policymakers in a structural bind: champion domestic technology on principle while users quietly rely on superior American tools in practice.
Austria’s proposal attempts to resolve that tension. If Anthropic, annoyed by its treatment in the US, were to establish a meaningful, legally grounded presence inside the EU while operating under European jurisdiction, values, and oversight, the bloc could potentially access frontier AI capabilities without formally abandoning its digital sovereignty commitments. It would not be US technology operating under US rules from US soil. It would be a company operating within the European legal framework.
What Comes Next
Gustavo Carriconde, Professor at Universidade Paulista, put the broader stakes clearly: “The latest reporting around Anthropic shows how quickly advanced AI can become entangled with export controls, national security concerns, and geopolitical dependency.”
For Carriconde, this signals a shift from much of the conversation in the EU about AI regulation:
“The key question for Europe is no longer only: Can we regulate AI? It is also: Can we ensure access to frontier AI infrastructure when access rules are controlled elsewhere?”
Carriconde noted that this does not mean Anthropic is moving to the EU, but that Europe is now openly debating whether hosting major AI capabilities inside the bloc should be treated as a strategic priority and that AI sovereignty is moving from theory to policy.
Whether the proposal advances will depend on how seriously the Commission engages with it and whether other member states align behind Austria’s position. With the EU’s AI Act and its Digital Decade targets both placing frontier capability at the center of European ambitions, political momentum is unlikely to dissipate, even if Anthropic’s appetite never materializes.
What the episode makes plain is that the era of passive dependency, in which European institutions simply accepted whatever access conditions American AI companies imposed, is drawing to a close.
