G7 weighs trusted partner access to advanced U.S. AI models

The G7 is weighing a new security club for artificial intelligence, one that would let select trusted partners into advanced U.S. models while Washington keeps everyone else out. The proposal would carve a narrow path for approved countries or companies to use frontier systems from firms such as Anthropic, turning access to AI into a test of alliance loyalty as much as technology policy.

The backdrop was Anthropic’s June 12 disclosure that the U.S. government had issued an export-control directive requiring the company to suspend all access to Fable 5 and Mythos 5 by any foreign national, whether inside or outside the United States, including foreign-national Anthropic employees. That order sharpened the stakes for the G7 talks in Evian-les-Bains, France, where leaders were not just debating commercial access but the reach of U.S. national-security controls over the most advanced models.

The discussions took place partly on the sidelines of the opening G7 summit dinner on June 15, with U.S. Commerce Secretary Howard Lutnick involved in the side talks. One source said the trusted partners category could mean either countries or companies, a sign that the policy is still fluid and that the line between alliance and market access has not been settled. Another source said the access idea was tied to a larger security goal: helping G7 members strengthen cyber defenses against rivals such as China.

That military and intelligence concern was central to Washington’s move. Lutnick said officials feared the models could be diverted to foreign military intelligence users in China, making frontier AI a matter of export control rather than simple innovation policy. A White House official said Trump’s team had an open line of communication with allies and remained committed to addressing national-security concerns with Anthropic’s models.

The broader significance reaches beyond one company or one summit. The G7, a bloc of seven major advanced economies, could set a template for which allies are trusted with frontier AI and which are left behind, redrawing the boundary between cooperation with friends and containment of rivals. If the talks advance, access to the most powerful U.S. models could become a strategic prize, with economic and military consequences far beyond Evian-les-Bains.