Von der Leyen presses G7 for Europe access to top AI models

Ursula von der Leyen used the G7 summit in Évian-les-Bains to argue that Europe’s access to the best AI models was in the mutual interest of the United States and the European Union. Her pitch landed at a moment when G7 leaders were already weighing a trusted partners framework that could give select countries or companies limited access to advanced American systems, including models from Anthropic.

The timing mattered. Anthropic disabled access to Fable 5 and Mythos 5 on June 12 after receiving a U.S. export-control directive that cited national security authorities. Anthropic said the restriction applied to any foreign national, whether inside or outside the United States. The move gave fresh urgency to a dispute that is no longer just about model performance, but about who gets to use the most capable systems and under what conditions.

That tension sits at the center of Europe’s broader dilemma. Brussels has spent years arguing for digital sovereignty, but its companies, researchers and governments still depend on American compute, chips and frontier models if they want to compete in cybersecurity, finance and industrial development. Von der Leyen praised U.S. efforts to make sure AI companies act responsibly when releasing powerful new systems, while pressing the case that the European Union should not be locked out of the strongest tools available.

Emmanuel Macron sharpened the business argument, saying he expected progress in the coming weeks on broadening access to leading U.S. AI models. His warning carried a commercial edge: nobody would buy U.S. AI if they feared it could be switched off at any moment. That concern has become part of the same debate as national security, because frontier models are now being treated as strategic infrastructure rather than ordinary software.

The summit also brought the issue into the same room as the companies shaping it. A working lunch gathered G7 leaders, outreach partners and tech executives including Donald Trump, Sam Altman, Demis Hassabis, Arthur Mensch, Aidan Gomez and Dario Amodei, with innovation and AI on the agenda. The leaders adopted joint statements at the close of the June 15-17 summit, but the real test now is whether the trusted partners idea becomes a concrete access mechanism. If it does, Europe could gain a narrower path to advanced U.S. models and a say in how they are tested and used. If it does not, Washington is likely to keep tighter restrictions in place while Brussels keeps pressing for inclusion it cannot yet secure on its own.