US to set AI chip export rules, tech giants to be gatekeepers – Reuters
The United States is preparing to introduce new regulations that would designate technology companies such as Alphabet (NASDAQ:GOOGL)’s Google and Microsoft (NASDAQ:MSFT) as global gatekeepers for the distribution of artificial intelligence (AI) chips, according to a report from Reuters, citing individuals with knowledge of the draft plan. These companies would be allowed to offer AI capabilities within their cloud services overseas without needing a license, provided they adhere to stringent reporting requirements and prevent Chinese entities from accessing these AI chips.
The forthcoming rules, details of which are being disclosed for the first time, are part of an effort by the Biden administration to balance the facilitation of AI chip exports with the prevention of their acquisition by entities that could pose a threat. There is a concern within the U.S. government that China could exploit AI technology to enhance its military capabilities, conduct advanced cyber attacks, or develop bioweapons.
While the Commerce Department has not commented on the specifics or the release schedule of the new regulations, sources have indicated that the administration’s plans could still be subject to change. Neither Google nor Microsoft has responded to inquiries regarding the measure.
This initiative follows a pattern set by a national security agreement signed in April between Microsoft and the U.S. government, allowing Microsoft to supply AI technology to the Emirati company G42, which has historical connections to China.
Under the draft rules, companies without gatekeeper status would still be able to apply for licenses to import a limited quantity of high-end AI chips from manufacturers such as Nvidia (NASDAQ:NVDA) and AMD (NASDAQ:AMD) in each country. Nvidia has expressed its willingness to collaborate with the U.S. government on the regulations, while AMD has not provided a response.
The proposed regime would exempt 19 allied countries, including the Netherlands, Japan, and Taiwan, allowing them unrestricted access to AI chips or the capabilities they enable. Conversely, countries under nuclear embargo, such as Russia, China, Iran, and Venezuela, would continue to be barred from obtaining U.S. AI semiconductors.
A final review of an “Artificial Intelligence Diffusion” rule is underway, as indicated by a government posting earlier this week, suggesting that the publication of the regulations could be imminent. This rule is believed to pertain to the AI chip export caps.
The Information Technology Industry Council, representing companies including AMD and Google, has voiced concerns that the Biden administration might be expediting the implementation of this complex rule without sufficient input from the industry, which could prevent unintended negative impacts. Naomi Wilson, the council’s senior vice president of Asia and global trade policy, stated that such changes could significantly alter the landscape of export controls with wide-ranging global effects.
These new rules expand upon a program introduced in September, which allows pre-approved international data centers to receive AI chips without a license. To qualify for this status, data centers are required to disclose extensive information about their operations, including customer details, business activities, access controls, and cybersecurity measures.
This article was generated with the support of AI and reviewed by an editor. For more information see our T&C.