Allies Want American AI, Minus the Kill Switch
The world wants American artificial intelligence. It just does not want Washington to be able to switch it off. That was the message delivered to President Trump at the Group of Seven summit, where French President Emmanuel Macron and Indian Prime Minister Narendra Modi warned, in pointed terms, that countries are uneasy about depending on U.S. AI systems they cannot control. Macron made the case over lunch with Trump and the chief executives of several American AI companies, arguing that AI supply security has become a geopolitical question with the same weight as energy security, according to officials who attended the session.
The concern is not about the quality of American models, which remain the most capable in the world and are widely adopted across allied economies. It is about the terms of access. The United States has used export controls and service restrictions to manage the spread of sensitive technology, and those tools have begun to reach AI itself. Companies have been ordered to stop offering certain models in specific markets, and at the G7 session one company disclosed that it had suspended service in parts of its overseas footprint under U.S. direction. For foreign governments, the message was unmistakable: the technology they depend on can be withdrawn at Washington’s discretion.
Macron’s framing drew a direct parallel to energy. European nations spent decades reducing their dependence on Russian gas, and the lesson of that episode, he argued, is that reliance on a single supplier is a strategic vulnerability even when the supplier is friendly. AI is becoming the energy of the digital economy, powering everything from government services to banking to defense, and Macron pressed the point that the countries adopting American systems need assurances they will not be cut off in a crisis. Indian officials made a similar argument privately, noting that their country’s huge technology sector has come to depend on U.S. models for everything from software development to customer service.
Trump’s response, according to people familiar with the exchange, emphasized American openness and the benefits of U.S. technology, while resisting commitments that would limit Washington’s discretion over sensitive systems. The tension is fundamental: the United States wants its AI companies to dominate global markets, but it also wants to preserve the ability to restrict those same technologies when national security is at stake. Those two goals are increasingly in conflict, and the G7 session put the conflict on display in front of the leaders and executives best positioned to resolve it.
For American AI companies, the stakes are commercial as well as strategic. International markets represent a growing share of revenue for the largest developers, and governments that fear being cut off will build alternatives. India, in particular, has become a heavy user of open-source models that can be hosted domestically, and several G7 attendees said the restrictions on American services have accelerated that shift. European nations are funding their own model development, and Japan and South Korea are backing domestic AI efforts with public money. None of those efforts matches American capability today, but they all reduce future dependence.
The lunch also touched on the rules of the road for AI exports. American officials have argued that controls are targeted at adversaries and should not affect allies, but allied leaders noted that the enforcement has not always distinguished cleanly between the two. A company that suspends service in one region may be serving an ally in the next, and the unpredictability of the policy is itself a problem, several officials said. They pressed for clearer criteria, advance notice and a consultation process before restrictions are imposed, requests that Washington has not yet embraced.
The episode reflects a broader realignment in how nations think about technology. For decades, the assumption was that technology flows freely among allies and that dependence was acceptable because the interests of the United States and its partners aligned. The AI era has complicated that assumption, because the same tools that give American companies their edge also give Washington an off switch. Allies are beginning to price that risk into their procurement decisions, and the companies themselves are caught in the middle, trying to serve global customers while complying with restrictions they often learn about late.
No joint statement on AI supply security emerged from the summit, according to officials, and none was expected. But the conversation at the lunch table may matter more than any communique. The leaders of two of the world’s largest democracies have now told the American president, in front of his own industry’s chief executives, that the price of AI dominance is trust. Whether Washington can square its security instincts with its allies’ demand for reliability will shape the global AI market for years, and for now neither side appears willing to blink. This article was prepared by Rhino Finance’s editorial team based on public reporting.
