Biden Administration Set to Introduce New Rule Limiting AI Chip Sales to China and Russia
WASHINGTON − On Monday, the Biden administration is likely to announce a new regulation concerning the export of sophisticated artificial intelligence chips. This move is intended to restrict access to advanced technology for countries like China, while facilitating acquisitions for 18 important allies and partners.
According to a statement from the White House, “Inappropriately handling powerful AI systems poses significant threats to national security, such as enabling the creation of weapons of mass destruction, supporting aggressive cyber operations, and facilitating human rights violations like mass surveillance.”
During a discussion with reporters on Sunday, Commerce Secretary Gina Raimondo emphasized that the new policy aims to foster a reliable global technology community.
She noted, “Addressing these significant national security concerns requires an understanding of the evolution of AI technology, our adversaries’ capabilities, and our allies’ interest in benefiting from this technology.”
The proposed rule will also allow flexibility, enabling countries aligned with U.S. interests to engage in substantial purchases.
No restrictions will be placed on chip sales to allies and partners, which include technological powerhouses like South Korea, Taiwan, various European nations, the U.K., and Australia.
Countries under arms embargoes such as China, Russia, and North Korea are already facing restrictions on exporting advanced AI chips. The new rule will impose additional limitations on transferring the most powerful AI models to these nations.
Orders for chips with collective computing power up to approximately 1,700 advanced GPUs will not require a license and will not count towards national chip limits, according to the White House. Most chip orders, especially those made by universities, healthcare institutions, and research organizations for benign purposes, fall into this category.
Raimondo remarked that the goal is to achieve “the dual objectives of allowing AI to expand globally, drawing other nations closer to us, while also ensuring our protection against national security threats.”
The rule, expected to be finalized during the last week of the Biden administration, will enter a 120-day comment period to gather input from experts, industry stakeholders, and partner nations.
Swapna Venugopal Ramaswamy is a White House Correspondent You can follow her on X @SwapnaVenugopal