The G7 is exploring a new security arrangement that would grant a limited group of trusted countries and companies access to cutting-edge U.S. artificial intelligence models, while barring others under stringent export-control measures. This approach reflects Washington’s effort to balance technological leadership with national security and alliance loyalty.
This initiative emerged after the U.S. government issued an export-control order requiring Anthropic, a leading AI company, to block all foreign nationals—including employees inside the U.S.—from accessing its most advanced AI models, Fable 5 and Mythos 5. The directive underscores concerns about AI technologies potentially being exploited by foreign military or intelligence agencies, particularly in China.
The discussions unfolded alongside the G7 summit in Evian-les-Bains, France, involving key figures such as U.S. Commerce Secretary Howard Lutnick. The proposed “trusted partner” category remains fluid, potentially covering both countries and corporate entities, signaling an ongoing debate over the intersection of geopolitical alliances and market access to frontier AI systems.
Central to the talks is a broader security objective: equipping G7 members to bolster cyber defenses amidst rising strategic competition from rival powers. Officials fear that unrestricted AI model access could aid adversaries in developing advanced military or intelligence capabilities, making control over AI exports a critical national security matter instead of merely an innovation policy issue.
While the current measures focus on one company and two specific models, the implications extend further. The G7’s stance could establish a global precedent for which allies earn privileged AI technology access and which nations remain excluded. This emerging “AI security club” reflects a shift toward using technology control as a tool of international alignment and strategic containment.