At the G7, Allies Split Over Who Governs AI — and Whether Anyone Should

While the cameras at Évian fix on three rival AI lab chiefs arriving together, the harder fight at this week's G7 is between the governments in the room — over who gets to govern artificial intelligence, and whether anyone should at all. Two years after these summits produced relatively hard commitments on the harms of generative AI, the 2026 edition, running 15 to 17 June, is expected to pivot toward the technology's economic upside, with any shared governance language watered down before the final communiqué is signed.

The United States has made clear it opposes multilateral agreements that could blunt its industrial lead. Washington's pitch is the "American AI technology stack" — exporting US hardware and software abroad, frequently backed by Commerce Department financing, as the preferred alternative to Chinese systems. The White House's science and technology office has reframed AI governance as a matter of national independence and sovereignty rather than of common, binding rules — a stance that leaves little room for the kind of collective pledge earlier G7 gatherings prized.

Europe is pulling the other way. On 3 June the European Commission unveiled its European Tech Sovereignty package: more EU-owned chip manufacturing and cloud capacity, and a deliberate push toward open-source to cut reliance on American giants. "We cannot afford to depend on others for the technologies that keep our hospitals running," Commission President Ursula von der Leyen said in laying out the plan — a line aimed as squarely at Silicon Valley as at Beijing.

Canada is carving out a third lane. Prime Minister Mark Carney unveiled a national AI strategy on 4 June built around public investment in domestic AI infrastructure and stronger international alliances. The moves are adding up: a Canada-Germany "Sovereign Tech Alliance" struck in February, Canadian firm Cohere's April acquisition of Germany's Aleph Alpha, an EU-South Korea pact on 10 June and an EU-Brazil digital partnership on 11 June — a widening web of arrangements that quietly route around the US stack rather than confront it head-on.

All of this collides at a summit already crowded with other crises: Ukraine, where President Zelenskyy is attending in the war's fifth year; the US-Iran framework and the reopening of the Strait of Hormuz; and a familiar argument over trade imbalances that France sums up as "China produces too much, the U.S. consumes too much and the Europeans invest too little." AI sits on the agenda alongside online child protection and digital infrastructure — but the deeper question, of who writes the rules for the decade's most consequential technology, is the one no joint statement out of Évian looks likely to answer.