The US government is introducing proposals to prevent foreign companies, particularly China, from using US cloud computing for AI model training, US Commerce Secretary Gina Raimondo announced this week, Reuters reported. The Biden administration sees this as an effort to protect national security and U.S. technological superiority. Meanwhile, Chinese companies can continue to access services deployed in Europe and the Middle East.
“You can’t involve non-state actors or China or people who don’t want access to the cloud to train models,” Raimondo said. Reuters. “We have export controls on chips,” she said. “Since those chips are in cloud data centers in the United States, we also have to think about blocking the path to potential malicious activity.”
A key aspect of the initiative is called “Know Your Customer,” which requires U.S. cloud companies to rigorously identify foreign users. The regulation would prevent companies from countries such as China from accessing U.S. cloud resources for artificial intelligence development. Raimondo compared this to existing export controls on high-performance AI processors (such as Nvidia’s H100) and the need to cut off potential avenues for malicious activity using American technology on U.S. soil. is emphasized. The initiative grew out of an October proposal to block Chinese companies’ access to U.S.-based cloud technology.
The proposed regulations impose significant responsibilities on cloud computing companies. These companies must verify the identity of foreign customers, maintain user identification standards, and certify compliance annually. This rule is part of a larger strategy to ensure that U.S. cloud platforms are not exploited for potentially hostile AI development. China, on the other hand, sees this as a way to curb economic development.
In part, U.S. regulation of certain entities accessing U.S. clouds is already in place. In October, President Joe Biden ordered developers of AI systems that may pose a threat to the national security, economy, public health, or safety of the United States to ensure that these systems are safe before making them available to the public. Enacted an executive order requiring test results to be disclosed to the U.S. government. public.
The industry’s response to these measures, at least in the public eye, has not been positive. Carl Szabo, general counsel for tech industry trade group NetChoice, criticized the executive order’s implementation as potentially illegal and could hinder international cooperation in the field of AI.