Features

AI CEOs join Trump and G7 leaders in Evian, pitch a U.S.-led standards forum

Sam Altman, Dario Amodei and Demis Hassabis sat with heads of state at a working lunch on the summit's final day, surfacing export controls, sovereignty and frontier-model risk as first-tier geopolitical issues.

By Markus Voskuhl · Features & longform · 2026-06-17 · 4 min read

On the final day of the G7 summit in Evian, Sam Altman sat between Donald Trump and Egyptian President Abdel Fattah el-Sisi at an hours-long working lunch, and according to an OpenAI official, he was the first chief executive to speak. The seating chart told the story before the communiqué did. Frontier-model CEOs are no longer lobbying governments from the outside; they’re now placed, literally, between them.

Altman was joined by Dario Amodei of Anthropic and Demis Hassabis of Google DeepMind, alongside about a dozen other tech leaders: Meta’s Alexandr Wang, Mistral’s Arthur Mensch, Cohere’s Aidan Gomez, Domyn’s Uljan Sharka, Synthesia’s Victor Riparbelli, Black Forest Labs’ Robin Rombach, and Salesforce’s Marc Benioff, with founders from India’s Sarvam and Japan’s Sakana also expected. On the government side: Trump, Emmanuel Macron, Keir Starmer, Friedrich Merz, and the U.S. economic-security triad of Treasury’s Scott Bessent, Commerce’s Howard Lutnick, and State’s Marco Rubio.

The pitch from the labs is a U.S.-led standards forum for democracies. Chris Lehane of OpenAI told reporters afterward that “countries and companies were coalescing around the idea of a space for democratic countries to develop standards that would help ensure ongoing access to frontier models.” Translation: a club, with membership criteria, whose gatekeepers are the same firms whose models are being gated.

The timing isn’t subtle. Export controls on Anthropic’s latest models went into effect late Friday, hours before Amodei walked into a room with the cabinet officials administering them. OpenAI had told CNBC earlier in June it expected “voluntary commitments” to emerge from Evian, a framing that pre-loads the outcome.

Two analysts read the moment in compatible ways. “The frontier labs want to shape this debate before any binding rules exist,” Emerson Brooking of the Atlantic Council told CNBC. Jessica Brandt of the Council on Foreign Relations noted that heads of state now need the cooperation, if not endorsement, of a handful of private-sector executives to make credible AI commitments.

That’s the structural fact. Sovereign AI rhetoric has converged with industrial policy and export control into something resembling the 1996 Telecommunications Act moment, where the regulated industry helped draft the rulebook governing it. The difference is that this time the drafting is happening at a lunch table, before anything is binding, and the export controls are already live.

Sources