During a meeting attended by technology leaders and heads of state, including U.S. President Trump, the CEOs of Anthropic and Alphabet's DeepMind unit called for the formation of a U.S.-led coalition to establish rules and standards for artificial intelligence.

The private luncheon took place on Wednesday in Évian-les-Bains, France, as part of the G7 summit. According to two individuals familiar with the matter, both Dario Amodei of Anthropic and Demis Hassabis of DeepMind proposed international cooperation on AI, with the United States taking a leading role, to mitigate the risks posed by the emerging technology.

As reported by one of the aforementioned sources and another person briefed on the discussions, Canadian Prime Minister Mark Carney concurred that the United States could spearhead such an AI alliance.

This meeting occurs as increasingly powerful AI models, possessing highly advanced cyber capabilities, are being released. Some industry experts fear these models could pose significant catastrophic risks if they fall into the wrong hands. Most recently, on Friday, Anthropic restricted access to its latest models, Fable 5 and Mythos 5, following the U.S. government's imposition of export controls on the new version, citing national security concerns.

In addition to Amodei and Hassabis, OpenAI CEO Sam Altman was among approximately a dozen technology executives present at Wednesday's gathering. Leaders from G7 nations also participated.

Representing the United States alongside President Trump were Treasury Secretary Scott Bessent, Commerce Secretary Howard Lutnick, and Secretary of State Marco Rubio.

One informed source stated that Amodei, in his remarks to attendees, suggested that areas for international collaboration should include orderly access to frontier models. The source added that Amodei emphasized the need for nations to cooperate in addressing AI risks related to cybersecurity, bioterrorism, and intelligence.

Anthropic remains in negotiations with the Trump administration following the enforcement of export controls on its latest models late Friday.

According to an OpenAI briefing, Altman advocated during his speech for "an international discussion forum to establish globally recognized testing standards, provide expert-level impartial analysis of capabilities and risks, and serve as a platform for cooperation among nations."

Last month, OpenAI announced that a variant of its latest model, GPT-5.5 Cyber, is being made available in a limited preview to vetted cybersecurity teams.

Chris Lehane, OpenAI's Global Head of Public Policy who attended Wednesday's meeting, noted that the non-U.S. leaders present acknowledged that the United States "can indeed play a leading role in advancing" the establishment of AI-related standards.