Anthropic’s Amodei, Google’s Hassabis call for U.S.-led AI coalition at G7, sources tell CNBC

In a pivotal meeting during the G7 summit in Évian-les-Bains, France, tech leaders including Anthropic's CEO Dario Amodei and Google DeepMind's co-founder Demis Hassabis advocated for a coalition spearheaded by the United States to develop regulations and standards for artificial intelligence. This private luncheon, attended by key figures such as President Donald Trump, highlighted the urgent need for international cooperation to mitigate the risks posed by rapidly advancing AI technologies. Sources familiar with the discussions revealed that both Amodei and Hassabis emphasized the importance of U.S. leadership in establishing a framework for AI safety. Canadian Prime Minister Mark Carney echoed their sentiments, agreeing that a U.S.-led initiative could help navigate the complexities of AI governance. Anthropic chose not to comment on the meeting, while representatives from Google DeepMind and the Canadian Prime Minister's office did not respond to inquiries. The call for collaboration comes in the wake of heightened concerns regarding the capabilities of emerging AI systems, particularly after Anthropic restricted access to its latest models, Fable 5 and Mythos 5, following U.S. export controls aimed at safeguarding national security. In addition to Amodei and Hassabis, OpenAI's CEO Sam Altman was present, along with several other tech executives and leaders from G7 nations. During the meeting, Amodei proposed that international efforts should focus on regulated access to cutting-edge AI models and the trade of crucial components, specifically excluding China to prevent potential risks. Amodei also highlighted the need for collective action on issues related to AI's impact on cybersecurity, bioterrorism, and intelligence operations. Altman reiterated the necessity for a global forum dedicated to establishing universally accepted AI testing standards and facilitating cooperation among nations. Furthermore, OpenAI's announcement last month regarding the limited preview of its GPT-5.5 Cyber model for select cybersecurity teams underscores the ongoing developments in AI capabilities. Chris Lehane, OpenAI's global affairs chief, noted that leaders from other nations acknowledged the United States' potential to take the lead in crafting AI standards, reinforcing the significance of this coalition initiative.