Chinese officials are worried that US frontier AI models like Anthropic's Mythos could be used as offensive cyber tools against China, according to people familiar with the matter.
There's an obvious tension in that fear: Anthropic won't let Chinese companies use its models, but China argues the company can't rule out those same capabilities being used to attack it. Anthropic has long pushed the US to tighten restrictions on advanced chips and AI, and last year it stopped providing services to Chinese-controlled companies.
China isn't preparing to act immediately. But if the US further restricts Chinese AI firms, Chinese authorities are studying response options — including sanctioning American AI companies or placing them on restricted lists.