Washington Locked Out Allies From AI — Then Asked for Their Cooperation at the G7

AI access hijacks the agenda

The formal G7 agenda listed frontier AI risks, youth safety, cyber threats, and bio-threats as discussion topics . But according to multiple sources, the access ban on Fable 5 and Mythos 5 overtook the planned program . European officials pressed the US to reconsider the directive, and concerns about US technological protectionism became the dominant theme .

Canadian Prime Minister Mark Carney publicly warned that the restrictions underscored the dangers of overreliance on a limited number of American AI providers . The European Commission simultaneously began examining the practical implications of the export controls, signaling a broader institutional concern about a weaponized US AI policy .

The 'trusted partners' proposal

Behind closed doors, US Commerce Secretary Howard Lutnick led discussions on a compromise: a "trusted partners" designation that would grant a limited number of allied nations or companies exemptions from the ban . Three diplomatic sources confirmed the talks took place during the opening dinner on June 15, but as of the summit's conclusion, no exemptions had been granted .

The concept, if formalized, would establish a new tier in US AI export governance — distinguishing between allied access and adversary access — setting a template for future model restrictions . But the lack of immediate action left G7 partners empty-handed.

France's Palantir decision

On June 16, the second day of the summit, French Prime Minister Sébastien Lecornu announced that France's domestic intelligence agency, the DGSI, would terminate its contract with US tech firm Palantir and replace it with French rival ChapsVision .

"We cannot accept new strategic dependencies in the digital sphere," Lecornu said in a video statement posted on X . The DGSI had renewed its contract with Palantir just six months earlier, making the reversal a direct and unmistakable response to the Anthropic export ban . Lecornu framed the move explicitly as a pursuit of "strategic autonomy" and announced a €655 million ($760 million) investment in AI .

The contradictory US ask

The core tension was structural: Washington was asking its closest allies to submit to US-led frameworks while simultaneously demonstrating that it would not hesitate to unilaterally cut off access to the most powerful technology they depend on.

The US pushed for AI governance language that promoted American economic benefit and competitive advantage, while continuing to block any binding multilateral rules on AI . At the same time, it was advancing a proposed minerals trading bloc that would use the Pentagon's "OPEN" AI initiative to set reference prices for critical minerals — an idea that met significant skepticism from G7 partners .

Allies questioned both the governance of an AI-driven pricing mechanism and the wisdom of handing Washington the pricing levers for an entire industry at the very moment the US was proving it would weaponize technology access .

China in the background

A recurring undercurrent throughout the summit was the risk that US AI protectionism would push allies toward Chinese alternatives. Canadian PM Carney's remarks, along with statements from multiple European officials, framed the pursuit of technological independence as an urgent priority . Though no G7 nation announced a shift toward Chinese AI models like DeepSeek during the summit, the possibility hung over the proceedings.

Where things stand

The G7 ended without a resolution. The "trusted partners" concept remained just that — a concept. No allied waivers were issued. France had publicly walked away from a major US defense contractor. And the US left the summit simultaneously promising AI leadership while demonstrating it was willing to lead alone.