The European Commission says it is assessing the fallout after a US export control directive forced Anthropic to disable its Fable 5 and Mythos 5 models worldwide on June 12. The order blocked access for any foreign national, and Anthropic shut the models down entirely rather than try to filter users by passport. European researchers have spent the days since arguing about what, if anything, Europe should do about it.
Brussels keeps it diplomatic
Commission spokesperson Thomas Regnier struck a careful tone, saying through the Commission that emergency measures should not be discriminatory toward partners and that this was a shared challenge, not one limited to a single company or jurisdiction. He also called it one more reason Europe needs to strengthen its own technological sovereignty. Which is the kind of thing officials say when an American agency has just demonstrated they can flip a switch and there is nothing Brussels can do about it.
The Commission is examining the practical consequences for European users. That is the polite version. The blunt version came from the researchers.
A wake-up call, but for what exactly
Thorsten Holz, scientific director at the Max Planck Institute for Security and Privacy, said what stood out was that a single foreign administrative act could switch off a model overnight for every non-US citizen, European companies and even Anthropic's own foreign staff included. His framing of digital sovereignty is worth sitting with: not self-sufficiency, but the ability to keep using critical technology even when geopolitics turns ugly. Holz laid this out in the Science Media Center's expert roundup, which is where most of the sharper European reaction landed.
Konrad Rieck of TU Berlin went further. American models can be turned off at any moment, he said, sometimes for opaque reasons, so Europe needs competitive models of its own and no way around it.
The Airbus pitch
Gitta Kutyniok of LMU Munich wants an "Airbus moment" for AI: pooled, ambitious investment in foundation models, chip design and energy-efficient computing. Build the structures now, she argues, because whoever waits until they exist will have little say in shaping them. It is a clean analogy. It is also worth remembering the actual Airbus took close to three decades to become a serious competitor, which is not the timeline anyone discussing frontier models has in mind.
"Europe will not be able to develop models like Mythos or Fable 5 in competition with the US." That is Paul Röttger of the Oxford Internet Institute, and he is the one voice in the room pouring cold water on the whole idea.
Röttger thinks pouring money into a homegrown rival is the wrong move. Better, he says, to lock in access through contracts tied to data-center investment and back it with credible trade-policy threats. He does not expect the Fable 5 ban to last, since it hurts US companies too much, but the signal is clear enough: access to American AI is not a given.
The energy problem nobody wants to mention
Matthias Hein of the University of Tübingen made the point that Europe needs not one domestic provider but several, because you cannot count on commercial firms to keep shipping open-weight models out of goodwill.
Then Jonas Geiping, also in Tübingen, did the math out loud. France's Mistral has fallen badly behind over the past two years. Closing the gap would need large-scale data-center construction and a real expansion of power generation, and here he dropped the detail that sticks: German electricity generation has fallen back to 1985 levels. You cannot train frontier models on nostalgia.
Geiping also pushed back on a comparison Anthropic itself likes to reach for, the parallel to nuclear weapons. AI is woven into the economy in a way warheads never were, he argued. Shutting it off or restricting it during a diplomatic spat would not just dent defense readiness; it could do serious damage to an economy whose processes increasingly cannot run without these systems.
Anthropic, for its part, has called the directive a misunderstanding and says it is working to restore access. The company laid out its case in a public statement, arguing the government acted on a narrow jailbreak that other available models, including OpenAI's GPT-5.5, are equally vulnerable to. Whether Washington agrees, and how fast, is the thing to watch.
