Anthropic CEO Dario Amodei and Google DeepMind CEO Demis Hassabis urged world leaders to form a U.S.-led coalition to guide artificial intelligence policy during a closed-door meeting at the G7 summit, according to people familiar with the discussion.

The meeting took place Wednesday in Évian-les-Bains, France, and included heads of state as well as about a dozen technology executives. President Donald Trump attended, along with senior members of his administration. OpenAI CEO Sam Altman was also among the executives at the gathering.

According to people with knowledge of the conversation, Amodei and Hassabis argued that countries should work together on AI rules and standards, with the United States taking the lead. Canadian Prime Minister Mark Carney reportedly said the U.S. could lead such an effort.

The discussion comes as governments and businesses grow more concerned about the security risks tied to increasingly advanced AI systems. Recent model releases have raised alarms because of stronger cyber capabilities, which some experts say could be misused if they fall into the wrong hands. Anthropic’s newest models, Fable 5 and Mythos 5, were recently disabled after the U.S. government imposed export controls on them over national security concerns.

Amodei reportedly said cooperation should focus on giving countries structured access to frontier AI models and on trade in chips and other critical components, while excluding China. He also pointed to risks involving cyber threats, bioterrorism and intelligence applications of AI.

Altman, through a briefing from OpenAI, called for an international forum that would set broadly accepted testing standards, provide independent analysis of AI capabilities and risks, and create a venue for cross-border cooperation. OpenAI recently began a limited preview of GPT-5.5 Cyber for vetted cybersecurity teams.

OpenAI’s global affairs chief Chris Lehane, who was also at the meeting, said participants from outside the United States acknowledged that the U.S. could play a leading role in setting AI standards.

The G7 discussion underscores how major AI companies are pressing governments to coordinate policy as the technology becomes more powerful and more widely deployed. It also reflects a growing effort by industry leaders to shape the rules before national approaches diverge further.

Anthropic declined to comment on the meeting. Google DeepMind and the office of the Canadian prime minister did not immediately respond to requests for comment.

The summit gathering followed a series of policy moves and company decisions that have highlighted the tension between innovation and security. Anthropic is still in talks with the Trump administration after export restrictions took effect on its latest models late last week.

With leaders from the world’s largest advanced economies gathered in one place, the meeting offered a rare forum for AI executives and governments to discuss how to coordinate on a technology that is increasingly central to economic and national security planning.