Anthropic and Europe: Testing Technological Sovereignty

The Halt of Anthropic: A Shockwave for European Sovereignty

The suspension of Anthropic's most advanced AI models, Fable 5 and Mythos 5, for non-American citizens has triggered an immediate reaction from the European Commission. This decision, made following an order from the U.S. government related to national security concerns, was deemed "discriminatory" by Thomas Regnier, spokesperson for the Commission on technological sovereignty. He emphasized that these emergency measures should not be discriminatory towards European partners and insisted on the need for Europe to strengthen its technological sovereignty.

The European Commission is currently assessing the practical impact of this U.S. export control order. According to Regnier, this development underscores the importance of a coordinated response to a challenge that is not limited to a single jurisdiction or company. The withdrawal of Anthropic's models has been seen as further evidence of the necessity for Europe to bolster its independent technological capabilities.

European Researchers Disagree on the Strategy to Adopt

European researchers have reacted differently to this situation, viewing it as a wake-up call. Thorsten Holz from the Max Planck Institute for Security and Privacy expressed concern about the ability of a single foreign government order to interrupt access to critical technologies for non-American citizens, including European businesses. He emphasized that digital sovereignty does not mean self-sufficiency, but rather the ability to use essential technologies even during periods of geopolitical tension.

Konrad Rieck from the Technical University of Berlin took a more direct approach. He warned that American models can be "turned off at any time, sometimes for opaque reasons," and stressed the need for Europe to develop and leverage its own high-performing models. In the same vein, Gitta Kutyniok from LMU Munich called for an "Airbus moment" for AI, advocating for a joint and ambitious investment in foundational models, chip design, and energy-efficient computing. She cautioned that anyone waiting for structures to already be in place will have little room to shape them.

On the other hand, Paul Röttger from the Oxford Internet Institute expressed a divergent opinion. He stated that he does not believe increased investment in European AI is the solution, asserting that Europe will not be able to develop models like Mythos or Fable 5 to compete with the United States. Röttger suggested that access should be secured through contracts linked to investments in data centers and supported by credible trade policy threats.

Structural Challenges for Europe

Two other researchers highlighted the difficulties of building local capacity. Matthias Hein from the University of Tübingen warned that Europe needs not just one, but several providers, as no one should rely on commercial companies to continue publishing open-weight models. Jonas Geiping from the ELLIS Institute in Tübingen pointed out the structural barriers behind this goal. He noted that the French company Mistral has "fallen significantly behind" over the past two years. Even though new players are emerging, the foundations are lacking: large-scale data centers and sufficient energy production, which in Germany has returned to levels of 1985.

Geiping also cautioned against historical parallels with tensions related to nuclear weapons, a comparison that Anthropic likes to make. Unlike nuclear weapons, he argued, AI is deeply integrated into the economy. A halt or restriction during a diplomatic conflict would not only affect defense capabilities but could also cause severe damage to the European economy if processes cannot function without high-performing AI.

Ongoing Discussions and Perspectives

Following these events, discussions are underway to restore access to Anthropic's models. Anthropic's CEO, Dario Amodei, will participate in a working dinner with G7 leaders on Wednesday, which could influence the course of negotiations and future policy decisions. Europe finds itself at a critical crossroads, faced with the choice of strengthening its technological autonomy or continuing to rely on foreign partners for essential technologies.