G7 leaders press US over Anthropic AI export ban and global rules

This G7 session is one of the clearest signals yet that frontier AI access is becoming a strategic asset on par with energy and semiconductors. Macron’s decision to publicly criticize Washington’s unilateral export curb on Anthropic’s latest models while simultaneously courting more AI investment in France underscores Europe’s twin anxieties: dependence on a handful of US labs, and falling behind in domestic capabilities.

For the labs, being at the table with heads of government formalizes a role they’ve effectively held for several years: de facto stewards of frontier model risk. Altman’s call for an international forum with standardized testing and impartial risk analysis is essentially a plea to move from ad hoc political directives—like the Anthropic ban—to predictable rule-making. The attendance list matters too: with OpenAI, Google DeepMind, Anthropic, Cohere, Mistral, Meta and others all present, we are watching the informal architecture of a ‘G7 AI bloc’ take shape.

In the race to AGI, this meeting doesn’t change the technical frontier overnight, but it does preview a world where cutting off model access can be used as leverage. That increases the incentive for Europe and US allies to build or host their own frontier-class systems, and for China to double down on sovereign models, accelerating a multipolar model ecosystem.