By David Shepardson

Eying China: WASHINGTON (Reuters) – The Biden administration is proposing requiring U.S. cloud companies to determine whether foreign entities are accessing U.S. data centers to train AI models, U.S. Commerce Secretary Gina Raimondo said on Friday.

“We can’t have non-state actors or China or folks who we don’t want accessing our cloud to train their models,” Raimondo said in an interview with Reuters. “We use export controls on chips,” she noted. “Those chips are in American cloud data centers so we also have to think about closing down that avenue for potential malicious activity.”

The Biden administration is taking a series of measures to prevent China from using U.S. technology for artificial intelligence, as the burgeoning sector raises security concerns.

The proposed “know your customer” regulation was released on Friday for public inspection and will be published on Monday. “It is a big deal,” Raimondo said.

The United States is “trying as hard as we can to deny China the compute power that they want to train their own (AI) models, but what good is that if they go around that to use our cloud to train their models?” she said.

Last month, Raimondo said Commerce would not allow Nvidia “to ship is the most sophisticated, highest-processing-power AI chips, which would enable China to train their frontier models.”

The U.S. government is worried about China developing advanced AI systems on a variety of national security grounds and has taken steps to stop Beijing from receiving cutting-edge U.S. technologies to strengthen its military.

The proposal would require U.S. cloud computing companies to verify the identity of foreign persons who sign up for or maintain accounts that utilize U.S. cloud computing through a “know-your-customer program or Customer Identification Program.” It would also set minimum standards for identifying foreign users and would require cloud computing firms to certify compliance annually.

Raimondo said U.S. cloud computing companies “should have the burden of knowing who their biggest customers are training the biggest models, and we’re trying to get that information. What will we do with that information? It depends on what we find.”

President Joe Biden in October signed an executive order requiring developers of AI systems that pose risks to U.S. national security, the economy, public health or safety to share the results of safety tests with the U.S. government before they are released to the public.

The Commerce Department plans to soon send those survey requests to companies. Raimondo told Reuters companies will have 30 days to respond. “Any company that doesn’t want to comply is a red flag for me,” she said.

Carl Szabo, general counsel at NetChoice, a tech industry trade group, said Commerce is implementing Biden’s “illegal” executive order “to force industry reporting requirements for AI.” He added that requiring U.S. cloud companies to report use of their resources by non-U.S. entities “for training large language models could deter international collaboration.”

Top cloud providers include Amazon.com’s AWS, Alphabet’s Google Cloud and Microsoft’s Azure unit.