Anthropic and Google DeepMind CEOs Push for US-Led AI Coalition at G7, Canada Expresses Support

At the G7 summit in Évian-les-Bains, Anthropic CEO Dario Amodei and Google DeepMind CEO Demis Hassabis used a closed-door lunch with heads of state and a dozen tech executives to call for a US-led coalition that would shape international rules and standards for frontier AI. Canadian Prime Minister Mark Carney reportedly agreed that Washington could lead such an effort. OpenAI CEO Sam Altman separately proposed an international forum for testing and standards. The gathering produced no binding commitments or regulatory announcements, but it signals how top AI labs are positioning themselves to influence governance even as governments pursue their own controls.

What happened

The meeting took place on the final day of the three-day summit. Amodei and Hassabis proposed a US-led AI coalition to establish international cooperation on structured access to frontier models, trade in chips and critical components excluding China, and addressing risks in cyber operations, bioterrorism, and intelligence. Canadian Prime Minister Mark Carney agreed the US could lead such a coalition, according to sources. OpenAI CEO Sam Altman called for a global testing forum to establish accepted standards for testing and serve as a cooperation venue. About a dozen tech executives attended, including Mistral's Arthur Mensch, Cohere's Aidan Gomez, Salesforce's Marc Benioff, and Meta's Alex Wang, alongside G7 leaders including President Donald Trump. The discussion occurred against the backdrop of US export controls on Anthropic's Fable 5 and Mythos 5 models, imposed just five days earlier due to national security concerns. No binding commitments or regulatory announcements emerged.

Why AI builders should care

For AI builders, this matters because the push for a US-led coalition could lead to standardized testing and access rules that affect which models are available, where they can be deployed, and what compliance obligations apply. The export controls on Anthropic's models show that governments are willing to act unilaterally, potentially disrupting products built on those models. If a coalition forms, it may harmonize some regulations, reducing fragmentation for builders serving global users. However, the lack of binding commitments means uncertainty remains. Builders should watch for concrete steps like structured access requirements or chip trade restrictions that could impact costs and infrastructure choices.

Practical implications

For now, the meeting is a conversation, not a negotiation. But the alignment of top AI labs behind a US-led framework suggests that future governance may favor the labs' own influence. Builders should consider diversifying model providers to reduce dependency on any single company's regulatory fate. OpenAI's proposal for an international testing forum could lead to standardized evaluation benchmarks, which might become de facto requirements for deploying models in regulated industries. The cyber dimension was prominent with OpenAI's GPT-5.5 Cyber limited preview and Anthropic's Mythos model restricted to defenders before export controls. Builders working on security products should monitor these developments.

Caveats

No binding commitments were made. The G7's track record on AI governance has yielded principles and codes of conduct but no enforceable regulation. The Trump administration has shown willingness to act unilaterally, which could undermine the collaborative framework. Several G7 members, including France and the EU, are pursuing their own AI regulatory paths that do not presuppose US leadership. The accounts of the meeting come from anonymous sources, and Anthropic declined to comment. The coalition idea may not materialize.

FAQs

What is the proposed US-led AI coalition discussed at the G7 summit?

Anthropic CEO Dario Amodei and Google DeepMind CEO Demis Hassabis proposed a US-led coalition to shape international rules and standards for frontier AI, focusing on structured access to models, chip trade excluding China, and cooperation on AI risks in cyber, bioterrorism, and intelligence.

Who spoke for the US-led AI coalition at the G7 and which companies were involved?

Dario Amodei and Demis Hassabis made the proposal. OpenAI CEO Sam Altman called for an international testing forum. About a dozen tech executives attended, including from Mistral, Cohere, Salesforce, Meta, Synthesia, Black Forest Labs, and others. Canadian Prime Minister Mark Carney supported US leadership.

Did the G7 meeting commit to any binding AI regulations or standards?

No. The gathering produced no binding commitments or regulatory announcements. It was described as a conversation rather than a negotiation.

What are frontier AI and why are standards for it being discussed?

Frontier AI refers to the most advanced models, often posing novel risks. Standards are being discussed because governments and labs disagree on how to manage these risks, and a US-led coalition aims to harmonize approaches to structured access, chip trade, and security cooperation.