OpenAI, DeepMind join Macron in G7 AI sharing push

This G7 episode is a marker of how central frontier AI has become to geopolitics. Macron is effectively telling Washington that treating Anthropic’s Fable and Mythos models as a strategic asset with an on/off switch doesn’t just punish one lab – it makes every allied economy dependent on U.S. goodwill for core digital infrastructure. By convening Altman, Hassabis and Amodei at the leaders’ table, France is elevating AI CEOs to the same level as heads of government, while also trying to socialise the risks of unilateral U.S. export decisions. ([apnews.com](https://apnews.com/article/7d783c6de4356962e338b8b8563d48ea?utm_source=openai))

For the race to AGI, the message is that access and control may matter as much as raw capability. If Europe, India and other G7 invitees fear a U.S. “kill switch”, they have strong incentives to build sovereign model stacks and data centers, potentially funding regional champions like Mistral, Cohere and others more aggressively. That could accelerate global R&D and diversification, but also fragment safety regimes as each bloc writes its own rules. The fact that Altman is publicly asking for an international forum on AI safety, even as his peers are pulled into national security disputes, underlines how unstable the current governance patchwork is. We are moving into a world where AGI labs must navigate not just technical scaling laws but also export controls, alliance politics and demands for shared stewardship. ([outlookindia.com](https://www.outlookindia.com/amp/story/international/g7-turns-focus-to-ai-sovereignty-as-concerns-grow-over-dependence-on-us-tech-giants?utm_source=openai))