Framework for Artificial Intelligence Diffusion
This interim final rule from the Department of Commerce's Bureau of Industry and Security revises Export Administration Regulations by tightening controls on advanced computing integrated circuits and introducing new export controls on the model weights of advanced closed-weight dual-use AI models. It establishes a global licensing framework alongside new license exceptions and validated end-user authorizations to facilitate responsible international diffusion while protecting U.S. national security. The regulation introduces specific compliance dates and reporting requirements for stakeholders handling advanced computing items and frontier AI model weights.
Summary and content fields are drawn from the document text and are pending editorial review.
Who issued it
Who it governs
Exporters, reexporters, AI developers, and cloud service providers must read this document to understand compliance obligations regarding the international transfer of advanced computing hardware and frontier AI models.
What the document contains
The document addresses export controls on advanced computing integrated circuits, semiconductor manufacturing items, and closed-weight artificial intelligence model weights. It establishes licensing requirements, license exceptions, validated end-user authorizations, compliance timelines, and specific security guidelines for data centers.
It does not provide ethical guidelines for AI fairness or bias mitigation, nor does it address domestic consumer privacy protections or local civil rights laws outside of export administration contexts.
- The rule establishes a worldwide license requirement for advanced computing integrated circuits and certain associated software and technology.
- New ECCN 4E091 introduces global export controls on the model weights of closed-weight AI models trained with 1026 or more computational operations.
- The regulation creates new license exceptions including AIA, ACM, and LPP to streamline low-risk transactions and manufacturing supply chains.
- Data Center Validated End-User authorizations are bifurcated into Universal (UVEU) and National (NVEU) categories subject to specific cumulative TPP caps and security standards.
- Open-weight models and models trained below the threshold are generally excluded from the new ECCN 4E091 controls.
· Background · Expanded Advanced Computing Integrated Circuit Controls · Overview of New Controls for AI Model Weights · Rulemaking Requirements · Regulatory Text Amendments (15 CFR Parts 732, 734, 740, 742, 744, 748, 750, 762, 772, and 774)
How it scores
Each score is backed by a quote verified against the document text. Editorial review has not yet confirmed them.
Accountability
4/5Clear responsibility for AI outcomes
The VEU will establish and bear responsibility for the following additional practices for AI security
Supplement No. 10 to Part 748, Section II, Paragraph 15
screening score · 2026-10-23
Fairness
0/5Non-discrimination and equitable treatment
screening score · 2026-10-23
Human Dignity
2/5Respect for the inherent worth of persons
assist in human rights violations, such as through mass surveillance
I. a. Risks and Benefits of Advanced AI Models
screening score · 2026-10-23
Human Oversight
3/5Humans remain in control of consequential decisions
ensure that the model weights of the most advanced U.S. AI models are stored outside the United States only under stringent security conditions
I. b. iii. Framework for the Diffusion of Advanced AI Technologies
screening score · 2026-10-23
Privacy
2/5Protection of personal data and private life
monitor, track, and surveil citizens, among other purposes, without regard for human rights
I. b. AI Diffusion Policy Review
screening score · 2026-10-23
Transparency
3/5Openness about how AI systems work and are used
ECRA also commands that BIS’s rules be ‘‘transparent, predictable, and timely,’’ and have ‘‘the flexibility to be adapted’’
I. Background
screening score · 2026-10-23
Enforcement
4/5Consequences and compliance mechanisms with real force
Failure to adhere to this agreement may result in the revocation of VEU status on a national or global basis, denial of chip allocations or export licenses, as well as other penalties as appropriate under U.S. laws and regulations.
Supplement No. 10 to Part 748, Section I, Paragraph 19
screening score · 2026-10-23
Impact Assessment
4/5Required assessment of an AI system's risks/impacts before or during use
Applicants are required to submit the total aggregated TPP volume of each export item on their license application.
II. D. 2. Information Required in License Application
screening score · 2026-10-23