Last Friday, Anthropic—one of the leading developers in the global artificial intelligence race—abruptly suspended access to its most advanced models, Fable 5 and Mythos 5, for all non-US citizens. The move followed a directive from the Trump administration citing national security concerns, effectively pulling a digital kill switch on users across Europe.
Across the continent, research labs, technology firms, and even hospitals had been piloting these US-made systems for core operations over the past year. Now they face a stark reality: their digital infrastructure can be unplugged overnight by a foreign government.
The incident has galvanized an unusual consensus among European political figures. Jordan Bardella, the far-right French presidential candidate, and Thomas Regnier, spokesperson for the European Commission, have both argued that nations failing to develop and control their own AI models will remain permanently dependent on external powers. This rare alignment underscores the depth of concern in Brussels and beyond.
Europe's Strategic Autonomy Push
Earlier this month, the European Commission unveiled an ambitious plan for European Strategic Autonomy in technology. Central to this is the new Cloud and AI Development Act, which aims to triple the continent's data centre capacity, construct massive "AI Gigafactories," and promote secure open-source alternatives to proprietary US models.
Yet Europe remains trapped in what analysts call a "compute gap." While the EU scrambles to build infrastructure, the sheer scale of investment from American giants dwarfs European efforts. Relying solely on moonshot projects may prove too slow to close the gap.
One alternative gaining traction is a coalition of "middle powers." By teaming up with nations such as the United Kingdom, Canada, and Japan, Europe could pool its fragmented computing resources, creating a kind of AI insurance policy against sudden restrictions. Such an alliance would not replace full sovereignty but could provide a buffer while European capacity matures.
The urgency is palpable. As the EU warns that US export controls on Anthropic's models must not discriminate against Europe, the continent's leaders recognize that without swift action, these kill switches will become a regular feature of Europe's digital future. The question is whether the political will and investment can match the rhetoric.
