Anthropic chief Dario Amodei used the G7 summit in France to push a simple message with a sharp edge: democratic governments should stop turning artificial intelligence into a diplomatic food fight. His warning came just days after the U.S. moved to restrict access to some Anthropic AI models for foreign citizens, a decision that forced the company to block the models for everyone, including users in the United States.
That makes this less a neat policy debate than a preview of the mess ahead. The AI race is already colliding with export controls, national security fears, and very real commercial damage, and the people building the systems are now telling governments to coordinate before they break the market they want to protect.
Anthropic’s warning to G7 leaders
According to Financial Times, Amodei urged leaders to ”resist the temptation to fall out over the spread of AI tools.” He delivered that line in front of Donald Trump, whose administration was behind the recent restrictions on Anthropic’s Mythos 5 and Fable 5 models for foreign citizens.
Anthropic could not build a clean geographic block, so it shut the models off for everyone. That is the awkward reality of AI controls right now: policymakers can order the gate shut, but the plumbing is still too crude to keep the right people in and the wrong people out.
Sam Altman backs wider AI access
OpenAI chief Sam Altman backed the broad thrust of Amodei’s appeal. He said AI tools for cyber defense should be available to all countries represented in the room, while also supporting the idea of technical cooperation among democratic states.
Demis Hassabis of Google DeepMind joined that camp as well, calling for an expert group led by the United States to handle these questions globally. The line from the industry is becoming harder to miss: companies want guardrails, but they do not want a patchwork of national bans that turn frontier models into political hostages.
Macron, Modi and Mistral flag the stakes
French President Emmanuel Macron used the summit to argue that the response around Anthropic showed how high the stakes are for the U.S. and its G7 partners. He said talks with Trump and other leaders were productive, but that the group still has not worked out how it should handle advanced AI models.
Macron called for tighter AI regulation, while warning that democracies risk weakening themselves if they stop cooperating. He said he wants a platform for discussion and coordination in the coming months so allied countries can settle on shared standards.
India’s Narendra Modi also voiced concern over the U.S. restrictions, saying democracies need access to AI models to protect critical infrastructure. Arthur Mensch, who leads French AI startup Mistral, focused on a different weak point: supply chains. His point was blunt – the hardware and infrastructure chain is so interwoven that hostile states can exploit it, and that makes coordination harder to dodge, not easier.
The next fight is likely to be less about whether governments should regulate AI than about who gets to decide the rules, and whether allied states can agree before they start locking each other out one model at a time.