The US government has unveiled an Interim Final Rule aimed at regulating the global distribution of advanced chips critical for artificial intelligence (AI). This policy, introduced by the Biden-Harris Administration and the Department of Commerce’s Bureau of Industry and Security (BIS), is designed to secure US leadership in AI technology while restricting access to adversarial nations. According to the White House, the rule simplifies export processes for trusted nations, strengthens security standards, and closes loopholes that could allow smuggling or misuse of advanced technology.
“Artificial intelligence is quickly becoming central to both security and economic strength,” the White House stated. “The United States must act decisively to lead this transition by ensuring that US technology undergirds global AI use and that adversaries cannot easily abuse advanced AI. In the wrong hands, powerful AI systems have the potential to exacerbate significant national security risks, including by enabling the development of weapons of mass destruction, supporting powerful offensive cyber operations, and aiding human rights abuses, such as mass surveillance.”
Updated frameworks for AI chip export regulations
The rule designates 18 allied and partner nations as exempt from restrictions on chip purchases, recognising their alignment with US foreign policy and security objectives. These jurisdictions are regarded as vital to fostering innovation and advancing AI development. To further facilitate exports, the regulation removes licensing requirements for smaller-scale orders, covering chip purchases with a combined computational capacity of up to 1,700 advanced GPUs. The White House has stated that this change is intended to benefit research institutions, universities, and healthcare organisations by expediting access to low-risk US technology for non-sensitive applications.
A significant feature of the new rule is the expansion of the Verified End User (VEU) programme, which introduces two categories of trusted buyers. Universal Verified End User (UVEU) status applies to organisations in allied nations meeting strict security protocols. This designation allows them to allocate up to 7% of their global computational resources across different countries, ensuring advanced training capabilities remain within the US and allied borders. National Verified End User (NVEU) status is available for organisations in trusted but non-allied nations, enabling access to up to 320,000 advanced GPUs over two years. This provision supports local and regional AI development while guarding against diversion risks. For entities outside the trusted framework, the rule imposes a cap of 50,000 advanced GPUs per country. This measure limits large-scale acquisitions while maintaining access for stakeholders engaged in critical applications, such as healthcare and government operations.
The rule introduces significant restrictions on closed-weight AI model weights trained on 1026 computational operations or more. These weights now require authorisation for export, reexport, or in-country transfer. Exceptions exist for deployments by US and allied entities, and open-weight models remain exempt from these controls. Enhanced security standards are also being implemented to safeguard sensitive AI technologies and prevent adversarial use. BIS has updated its Data Center Validated End User (VEU) programme to support US technology’s secure diffusion. Universal VEUs (UVEUs) must retain at least 75% of their chips within the US and allied nations, while National VEUs (NVEUs) are authorised to build data centres under defined conditions. Caps on chip allocations ensure exports align with US security interests, with arms-embargoed countries facing a presumption of denial for all applications.
The regulation encourages government-to-government agreements with like-minded nations, allowing them to double their chip caps to 100,000 GPUs. This provision fosters international collaboration while ensuring compliance with US export controls and security standards.
“This policy will help build a trusted technology ecosystem around the world and allow us to protect against the national security risks associated with AI, while ensuring controls do not stifle innovation or US technological leadership,” said US Secretary of Commerce Gina Raimondo.
The new rule builds on earlier chip control regulations introduced in October 2022 and October 2023, which targeted China’s access to advanced semiconductor technologies. These previous measures focused on limiting adversarial nations’ ability to train advanced AI systems. The updated rule represents an evolution of those policies, incorporating insights from consultations with industry leaders, bipartisan congressional representatives, and allied nations over the past 10 months.
Read more: China vows retaliation if US enacts tighter semiconductor restrictions
