G7 AI Summit in Évian-les-Bains: Frontier AI CEOs Enter Geopolitical Room

The 52nd G7 Summit in Évian-les-Bains, France (June 15–17, 2026) included a working lunch where the CEOs of OpenAI, Anthropic, Google DeepMind, and Mistral sat alongside G7 heads of state — a first for a summit at this level [4][13]. The central AI policy debate turned on whether allied democracies should build shared AI infrastructure or fragment into rival national systems. Dario Amodei (Anthropic), Sam Altman (OpenAI), and Demis Hassabis (Google DeepMind) argued for the former, calling for shared cyberdefense tools, testing infrastructure, and technical standards [5]. European governments were reported as cautious about being drawn into an existing U.S.–Europe dispute over Anthropic's export controls [9].

Frontier AI company leaders now have a formal seat in rooms where consequential geopolitical decisions are made — not as invited experts but as participants in the same session as heads of state. The allied-AI-coordination question has moved from think-tank reports to live diplomatic negotiation, with real disagreements between the U.S. and Europe about how far that coordination should go.

The 52nd G7 Summit convened in Évian-les-Bains, France on June 15–17, 2026, under French presidency. Leaders of the seven member nations were joined by invited guests including India's PM Narendra Modi and Kenya, the sole African nation present [1][2]. AI governance sat alongside Ukraine, Middle East developments, and trade on the agenda. A preparatory workshop on frontier AI models' cyber capabilities and governance was held at Sciences Po in Paris in the days before the main summit [3].

The summit's most structurally significant AI moment was a working lunch that brought frontier AI company CEOs into the same session as G7 heads of state. Named participants included Sam Altman (OpenAI), Dario Amodei (Anthropic), Demis Hassabis (Google DeepMind), Arthur Mensch (Mistral), and several other executives — Aidan Gomez, Uljan Sharka, Victor Riparbelli, Robin Rombach, Alex Wang, and Marc Benioff [4]. The inclusion was widely noted as evidence that AI has been elevated to a peer-level geopolitical issue rather than a technical advisory matter.

The substantive position advanced by the AI CEOs centered on allied coordination. Amodei argued that democratic allied nations should pursue shared AI access rather than develop separate rival systems; Altman and Hassabis explicitly backed the same position, calling specifically for shared cyberdefense tools, shared testing infrastructure, and common technical standards [5]. Amodei and Hassabis were also reported to have urged G7 leaders to create a U.S.-led international AI framework [6]. Amodei separately warned against democratic nations splitting on AI governance, framing fragmentation as a strategic risk [7]. Canadian PM Carney had already invoked the Anthropic export control argument in a June 14 speech in Ireland, signaling the issue was on the pre-summit political agenda [8].

European governments took a more guarded posture. A Politico report described Europe as wary of openly stoking what it characterized as an 'Anthropic row' at the G7 [9], suggesting the export control dispute between the U.S. and European parties is live enough to create diplomatic friction at the table. The summit also took place in the wake of France's June 1 Choose France summit, which had produced €93bn in announced AI investment commitments including SoftBank participation [10], and against a broader discussion of AI access for 'trusted partners' [11]. G7 Leaders' Joint Statements were issued on June 16–17 [12], but detailed AI-specific outcomes from those documents have not yet been widely covered.