The 52nd G7 summit closed today in Évian-les-Bains with artificial intelligence dominating the agenda, as heads of state shared the stage with the founders of the world’s leading AI labs. Governance, child safety online, and the widening gap between model capability and regulation framed the day’s most consequential developments.
G7 agrees first joint approach to child safety and “safe AI growth”
Leaders from Canada, France, Germany, Italy, Japan, the UK, the US and the EU endorsed a common set of principles requiring online services to build safety and effective age assurance into products by design. The framework also commits members to address risks from AI chatbots and to improve data sharing among platforms, parents and researchers. The UK government called it the bloc’s first-ever joint approach to protecting children online while supporting responsible AI deployment.
AI’s top founders lobby leaders in person
For the first time, the chief executives of OpenAI, Anthropic and Google DeepMind — Sam Altman, Dario Amodei and Demis Hassabis — attended a G7 summit together, alongside Mistral’s Arthur Mensch, Cohere’s Aidan Gomez and Meta’s Alexandr Wang. The tech contingent focused on AI infrastructure and regulation, the two pressure points where frontier labs and governments keep colliding as capability races ahead of the rules meant to contain it.
OpenAI presses for a youth AI safety institute
Ahead of the summit, OpenAI published a set of youth safety principles — privacy-preserving age estimation, default safeguards when a user’s age is uncertain, annual risk assessments and stronger protocols around self-harm and exploitation. The company proposed a dedicated youth AI safety institute to keep standards aligned with fast-moving systems beyond any single summit.
Europe’s labs ride the policy moment
Mensch’s presence underscored Europe’s push to field a home-grown frontier champion. Mistral has continued to expand its ambitions against OpenAI and Anthropic, and the summit gave European players a platform to argue that sovereignty and safety can advance together rather than at odds.
What to watch: whether this week’s voluntary commitments gain enforcement teeth — and whether Google ships the long-expected Gemini 3.5 Pro, which Sundar Pichai has promised “next month,” before the policy ink dries.