Anthropic and Google's DeepMind bosses want an AI club with no China

At the G7 summit in Evian-les-Bains, the CEOs of Anthropic and Google DeepMind pitched a US-led AI governance coalition designed to leave China on the outside looking in. Dario Amodei argued for a system where the most powerful models, high-end chips, and critical hardware components are traded exclusively within a friendly, non-Chinese ecosystem. The plan also includes joint efforts to curb risks in cybersecurity, biological threats, and intelligence gathering.

While Demis Hassabis backed the coalition idea, the specific push to lock out China came directly from Amodei. The ironies were thick: Amodei was pushing this vision while sitting next to Howard Lutnick, the US Commerce Secretary who, just five days earlier, had forced Anthropic to disable its own top-tier Fable 5 and Mythos 5 models due to export controls. Donald Trump, the man whose administration signed off on that digital leash, was also at the table.

Not everyone was sold on the exclusionary club. OpenAI’s Sam Altman countered with a proposal for a neutral international forum, emphasizing that AI safety shouldn't be left to companies alone. Meanwhile, Emmanuel Macron openly mocked the approach, labeling the US move as strictly nationalistic and admitting he had been begging Trump to lift the restrictions on Anthropic's tech.

The coalition currently exists more as a wish-list than a policy, highlighting a desperate power play by labs to set the rules before governments do it for them. It is adorable that companies think they can dictate geopolitical borders while struggling to keep their own products online, proving once again that tech giants prefer a world where they are the ones holding the whip.

Source: CNBC