At the G7 Summit on Wednesday, some of the world’s most powerful leaders said the quiet part out loud: they depend on American AI, and that dependence scares them. According to TechCrunch AI, French President Emmanuel Macron and Indian Prime Minister Narendra Modi told fellow leaders and top AI executives that the U.S. could cut off access to leading American models at any moment. The warning landed in a room that included Anthropic CEO Dario Amodei, OpenAI CEO Sam Altman, and President Donald Trump.
Macron put it bluntly over lunch. If the U.S. “from one day to the next can turn off the switch,” he said, it would hurt European customers and damage the AI firms themselves. What stands out here is who was making the argument: not critics of American tech, but its biggest customers.
What triggered the alarm
The timing isn’t an accident. TechCrunch AI reports that just days earlier, the Trump administration blocked Anthropic from exporting its newest Mythos 5 and Fable 5 models on national security grounds. The order followed a flag from Amazon to the White House that certain safety guardrails could be bypassed.
Here’s the part that worries everyone outside Washington. Cybersecurity experts have pointed out that the same capabilities the government cited also exist in models that stay freely available, including from OpenAI. Anthropic’s models are still frozen anyway. The lesson international buyers took away: access can vanish overnight, for reasons you may never be told.
Why this matters
For years, the pitch for U.S. AI was simple. It’s the best, so build on it. That pitch now carries a hidden cost.
- Any company or government building on American AI infrastructure has to plan for sudden revocation.
- A startup in Paris or Bangalore could see its product break without warning, with no appeal.
- “Digital sovereignty” stops being a talking point and becomes a procurement question.
Modi said he shared the concern about the Anthropic block, according to Financial Times reporting cited by TechCrunch AI. He argued democratic nations need unfettered access to top models to protect critical infrastructure. Aidan Gomez, co-founder and CEO of Canadian firm Cohere, was sharper: “companies and democratic nations remaining dependent on a small handful of big tech companies is dangerous to resilience.” He framed the fight as being about “who controls the foundational technology that will shape our economic security and national sovereignty for decades to come.”
The proposed fix: ‘trusted partners’
G7 leaders floated a workaround. The idea is a “trusted partners” scheme that would grant non-U.S. nations reliable access to advanced models from firms like Anthropic and OpenAI, creating an open trade network that routes around U.S. restrictions. Both countries and companies could qualify, as long as they used the models to build stronger defenses against rivals like China.
The catch is that nobody knows how far it would reach. A government-to-government pact does little for a small company that just watched its product go dark. Macron’s pitch to Washington was practical: back the scheme, broaden Mythos access, because nobody wants to buy AI that could disappear overnight.
What to watch next
This is significant because it reframes the AI sovereignty debate. Europe and others have spent two years arguing they should build their own models. That case gets harder every time American systems pull further ahead and nobody wants to be left out. So the new strategy isn’t to replace U.S. AI. It’s to lock in guaranteed access to it.
Expect three things to develop: pressure on Washington to formalize export rules so allies aren’t caught off guard, renewed funding for sovereign and open alternatives as a backup, and enterprise buyers asking vendors hard questions about continuity clauses. The off switch is now part of the buying decision. More details are available in the original TechCrunch AI report.