G7 leaders demand AI access without U.S. shutdown risk

At the G7 Summit, world leaders expressed concerns over the U.S.'s ability to abruptly cut off access to advanced AI models, citing recent restrictions on Anthropic's Mythos 5 and Fable 5 as a cautionary example. French President Emmanuel Macron and Indian Prime Minister Narendra Modi emphasized that such unilateral moves could destabilize economies and undermine trust in U.S. AI infrastructure. Macron warned that if the U.S. can 'turn off the switch' without warning, it risks harming both European customers and American AI firms themselves.

The episode has exposed a broader geopolitical risk: countries and companies relying on U.S. AI models now face the possibility of overnight access revocations, even for models with comparable capabilities already available. The U.S. blocked Anthropic's models over alleged safety guardrail bypasses flagged by Amazon, though experts argue similar risks exist in unrestricted models. In response, G7 leaders discussed a 'trusted partners' scheme to grant non-U.S. nations access to advanced AI models while bypassing restrictions, but details remain unclear. The debate underscores a growing demand for digital sovereignty, with leaders like Cohere CEO Aidan Gomez warning that dependence on a handful of U.S. tech firms poses long-term risks to economic and national security.