Now loading...
Dario Amodei, CEO of Anthropic, and Demis Hassabis, CEO of Google DeepMind, urged for a coalition led by the United States to establish standards and regulations regarding artificial intelligence during a private meeting with technology leaders and government officials, including President Donald Trump, at the G7 summit in Évian-les-Bains, France. The discussions, which took place on Wednesday, focused on the need for international collaboration to mitigate the risks associated with advanced AI technologies, according to sources familiar with the matter who opted to remain anonymous.
Canadian Prime Minister Mark Carney acknowledged the potential for the U.S. to spearhead an AI coalition, aligning with the sentiments expressed by Amodei and Hassabis. Neither Anthropic nor Google DeepMind has commented on the discussions, and inquiries to the Office of the Prime Minister of Canada have gone unanswered.
The meeting comes amid rising concerns about the safety of powerful AI models, particularly following the recent introduction of cutting-edge tech with significant cyber capabilities that could lead to catastrophic outcomes if mismanaged. In response to national security worries, Anthropic recently restricted access to its latest AI models, Fable 5 and Mythos 5, following U.S. government export controls.
Attendees of the meeting included OpenAI CEO Sam Altman and several prominent executives from the tech industry, alongside leaders from G7 nations. The U.S. delegation was represented by officials such as Secretary of the Treasury Scott Bessent and Secretary of State Marco Rubio.
During his remarks, Amodei highlighted the importance of structured access to cutting-edge AI models and the need for cooperation regarding the trade of critical components, specifically suggesting exclusions for China. He also emphasized the necessity to address the risks associated with AI in areas such as cybersecurity and bioterrorism.
Anthropic is currently engaged in negotiations with the Trump administration following the introduction of export restrictions on its latest models. Altman advocated for the creation of an international platform to establish universally accepted testing standards and facilitate cooperation between nations on AI-related matters. OpenAI had recently announced the limited rollout of GPT-5.5 Cyber, a new version of its model designed for cybersecurity teams.
Chris Lehane, OpenAI’s global affairs chief who participated in the meeting, noted that leaders from other countries recognized the potential role of the U.S. in leading the establishment of AI standards.
