Anthropic Chief Executive Dario Amodei and Google DeepMind Chief Executive Demis Hassabis used a closed-door lunch with President Trump and other G7 leaders Wednesday in Evian-les-Bains, France, to press for a U.S.-led coalition that would set global rules for artificial intelligence, CNBC reported, citing people with knowledge of the meeting. Canadian Prime Minister Mark Carney told the room he agreed the United States could lead such a coalition.
The pitch lands at an awkward moment for Amodei. Anthropic remains in talks with the Trump administration after a federal export-control directive forced the company on Friday to disable its Fable 5 and Mythos 5 models for every customer, foreign and domestic. Wednesday's intervention is the most public attempt yet by Anthropic's chief to argue that the U.S. should be writing AI's international rules while he is asking Washington to relax the ones just imposed on his company.
What Amodei asked for
Amodei told the group that international cooperation should include structured access to frontier models and trade in chips and critical components that excludes China, according to one person in the room. He also called for cooperation against AI-enabled cyber, bioterrorism and intelligence risks. OpenAI Chief Executive Sam Altman, who attended alongside about a dozen tech executives, urged "an international forum for discussion that establishes globally accepted standards for testing, provides expert and impartial analysis of capabilities and risks, and serves as a venue for cooperation among nations," according to a briefing from OpenAI.
Treasury Secretary Scott Bessent, Commerce Secretary Howard Lutnik and Secretary of State Marco Rubio sat with Trump on the U.S. side. OpenAI global affairs chief Chris Lehane said non-U.S. leaders acknowledged that Washington "certainly could play the lead role in working to establish" AI standards.
India's reckoning
The export order is already reshaping strategy in New Delhi, which had bet on building AI applications atop foreign foundation models rather than its own. Saket Dandotia, co-founder of Onetab.ai, told CNBC that "[t]he fact that frontier access can vanish overnight on a foreign government's order is the whole problem." Dandotia said he survived Anthropic's cutoff only because he had spread workloads across multiple vendors, adding that "diversification buys time; it doesn't buy independence."
An ADP Research report released Thursday found 41 percent of Indian workers use AI nearly every day, against 26 percent in China and 19 percent in the United States, despite India producing no cutting-edge chips and no frontier-scale model of its own. Sarvam AI, the country's leading sovereign-model startup, raised $300 million Monday at a $1.5 billion valuation; its flagship model carries a little over 100 billion parameters. Venture investor Mohandas Pai has urged Prime Minister Narendra Modi to launch a new AI mission, calling existing programs "too slow, way too small" to matter.
The counterpoint
The Trump administration imposed the Friday curbs citing national security concerns, according to CNBC, and the White House and Commerce Department had not publicly addressed the G7 lunch or Amodei's coalition pitch by press time. Prediction-market traders are betting the dispute resolves quickly: Kalshi contracts give 58 percent odds U.S. access to Fable 5 is restored by July 1 and 74 percent by July 10, with Polymarket at 67 percent for July 1.
The next test is bilateral, not multilateral. Anthropic's talks with the Trump administration continue this week in Washington.