World leaders want American AI. They just don’t want America to be able to turn it off.

G7 leaders warned Wednesday that the United States could sever access to advanced American artificial intelligence models overnight, a concern sharpened by the Trump administration’s recent block on exports of Anthropic’s newest systems on national security grounds. French President Emmanuel Macron told G7 leaders and AI executives — including Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, and President Donald Trump — that if the U.S. “from one day to the next can turn off the switch,” it could harm European economies and damage the AI firms themselves [1]. The remarks came days after the White House blocked Anthropic from exporting its Mythos 5 and Fable 5 models after Amazon flagged that certain safety guardrails could be bypassed [1]. Cybersecurity experts have noted that the capabilities cited by the government also exist in models that remain freely available, including from OpenAI, yet Anthropic’s systems remain restricted [1]. The episode has exposed a structural vulnerability for international companies and governments building on U.S. AI infrastructure: access can be revoked overnight without explanation [1]. Indian Prime Minister Narendra Modi expressed concern about the Anthropic block, saying democratic nations must have unfettered access to top AI models to protect critical infrastructure [1]. Aidan Gomez, co-founder and CEO of Canadian enterprise AI firm Cohere, said in a statement that the restriction “confirms what we at Cohere have known all along: that companies and democratic nations remaining dependent on a small handful of big tech companies is dangerous to resilience” [1]. He added that digital sovereignty concerns “who controls the foundational technology that will shape our economic security and national sovereignty for decades to come” [1]. The dependency risk is amplified by the concentration of frontier AI development. The current AI boom, sometimes called an AI spring, has been driven by large language models and generative AI technologies developed by a small group of firms including OpenAI, Google, and Anthropic [2]. Researchers including Geoffrey Hinton, Yoshua Bengio, and AI CEOs such as Dario Amodei and Sam Altman have voiced concerns about the implications of advanced AI systems [6]. In 2023, hundreds of AI experts signed a statement declaring that mitigating the risk of extinction from AI should be a global priority alongside pandemics and nuclear war [6]. During the summit, G7 leaders discussed creating a “trusted partners” scheme that would grant non-U.S. nations access to advanced AI models from firms like Anthropic and OpenAI [1]. The goal is to maintain an open trade network that bypasses U.S. restrictions, with both countries and companies eligible as trusted partners if they use the models to develop stronger defenses against rivals like China [1]. Macron argued it would make sense for Washington to back such a scheme and to ensure broader Mythos access, noting that nobody would want to buy U.S. AI access if it could disappear overnight [1]. The discussions unfolded as Europe and other non-U.S. countries push for AI sovereignty, a case made harder when American models keep pulling ahead [1]. The scope of any trusted-partner framework remains unclear, as does whether it would protect a startup in Paris or Bangalore whose product breaks without warning [1].