Anthropic CEO urges G7 to align on AI after US export block

In short: Anthropic CEO Dario Amodei asked G7 leaders to stay aligned on AI policy after the US blocked exports of Anthropic’s newest AI models.

What happened

Anthropic chief executive Dario Amodei told leaders of the G7 group of major democracies that they should “resist the temptation to splinter” on advanced AI tools. He spoke at a working lunch in Évian-les-Bains, France, in front of US President Donald Trump and other leaders, according to people familiar with the discussion.

Amodei said he understood why countries want to stop AI from “getting into the wrong hands,” but argued that democratic countries should work together. His message was supported by Sam Altman, the chief executive of rival AI company OpenAI, who said cyberdefence tools should be shared with all countries in the room.

Demis Hassabis of Google DeepMind also attended. The three executives urged US-led cooperation on developing and testing advanced AI systems. They also raised risks like bioterrorism and cyber security, especially if allies cannot agree on shared rules and checks.

The comments came days after the US blocked Anthropic from exporting its Mythos and Fable AI models, citing national security. The Trump administration reportedly received information suggesting the models’ safety controls could be bypassed (like a lock that can be picked).

French President Emmanuel Macron said the dispute showed what is at stake if the US can “turn off the switch” for allies who rely on American AI. India’s Prime Minister Narendra Modi also raised concerns, saying democracies need access to AI to protect critical infrastructure.

Why it matters

If countries cannot access the same AI tools, it could affect business, public services, and security cooperation. The debate is also about trust, whether allies can rely on each other for key technology, like relying on a shared power grid.

Source: Financial Times