The US government has, for the first time, directly restricted the export of AI algorithms, intercepting Anthropic's most advanced models — Claude Fable 5 and Claude Mythos 5 — before they leave the country. Unable to verify user nationality in real time on the cloud, Anthropic had no choice but to pull the models globally. The sudden ban sent shockwaves through allied nations in Europe and India, who suddenly realized that their digital lifelines are held in Washington’s hands — and could be cut with a single command.
That panic erupted into a full-blown confrontation at a G7 summit lunch on June 17 in France, as world leaders and AI executives argued face-to-face with President Trump. In an attempt to salvage global markets, the tech giants pitched competing geopolitical strategies. Anthropic founder Dario Amodei and DeepMind CEO Demis Hassabis proposed a US-led, exclusionary AI alliance that would lock China out entirely by restricting frontier models and chip hardware. OpenAI CEO Sam Altman pushed back, advocating for a neutral, open forum — a way to keep cooperation alive even as Cold War tensions escalate.
The lunch was a hotbed of disagreement, with every leader nursing their own concerns. Indian Prime Minister Narendra Modi protested that without access to the strongest models, his country's infrastructure defenses would be left naked. French President Emmanuel Macron warned the Trump administration that cutting off allies would backfire, ultimately hobbling the trillion-dollar global ecosystems built by American tech giants. While G7 leaders discussed a trusted partner exemption mechanism that would allow security-cleared allied companies onto a special whitelist, the message was clear: control is rapidly consolidating in Washington. Allies either submit to US security assessments or risk being locked out of the world's most advanced productive technologies.