AI Diffusion Rule
In plain wordsA 2025 US rule, cancelled before it took effect, that would have sorted the world's countries into three groups with different limits on how many advanced AI chips they could buy.
An interim final rule issued by the US Bureau of Industry and Security on January 13, 2025, in the last week of the Biden administration, to govern the worldwide spread of advanced AI computing. It sorted destinations into three tiers: the United States and 18 close allies and partners (Tier 1) could receive advanced chips without restriction; arms-embargoed countries including China and Russia (Tier 3) were effectively barred; and most other countries (Tier 2) faced caps on total computing power, which companies could exceed by qualifying as 'validated end users' meeting security standards. Small orders, below roughly 1,700 H100-equivalents, were exempt. It also created the first export control on AI model weightsweights, n.: The learned numbers inside a neural network that set how strongly each part influences the next. The finished set of weights is, in effect, the trained model itself. (ECCN 4E091), covering closed models trained with more than 10²⁶ operations while exempting open-weight modelsopen-weight model, n.: An AI model whose trained internal numbers are published for anyone to download and run on their own computers, even if the data and code used to make it stay private.. Chipmakers, cloud providers and many Tier 2 governments objected, and on May 13, 2025, two days before its May 15 compliance date, the Trump administration announced that it would not enforce the rule and would rescind it.
Under the AI Diffusion Rule, a data center in a Tier 2 country would have needed validated-end-user status to exceed its cap.
Origin
Developed by the Commerce Department under Secretary Gina Raimondo and published in the Federal Register in January 2025; its rescission was announced on May 13, 2025 together with guidance that using Huawei's Ascend chips anywhere risked violating US export controls.
Usage note
Its successor policy relied on bilateral arrangements, notably large chip and data-center agreements with the United Arab Emirates and Saudi Arabia in May 2025, rather than a global tier system. It remains the most ambitious attempt so far to apply compute governancecompute governance, n.: Using control over the computer chips and data centers that AI depends on — who can buy them, where they go, how they are used — as a way to oversee powerful AI. worldwide.
Related
- export controls on AI chipsexport controls on AI chips, n.: Government rules that limit which countries and companies may buy the most advanced AI chips and the machines used to make them. The United States uses them mainly to slow China's progress in AI.
- compute governancecompute governance, n.: Using control over the computer chips and data centers that AI depends on — who can buy them, where they go, how they are used — as a way to oversee powerful AI.
- compute thresholdcompute threshold, n.: A legal cutoff based on how much computation was used to train an AI model. Models trained with more than that amount face extra rules, on the theory that bigger training runs produce more capable models.
- know-your-customer for computeknow-your-customer for compute, n.: Rules that would make cloud companies check who their customers are before renting them enough computing power to train a large AI model, much as banks must verify who opens an account.
- export control directiveexport control directive, n.: A US government order, used against AI models in June 2026, telling a company to stop foreigners from using some of its AI models for national-security reasons.
- sovereign AIsovereign AI, n.: A country's ability to build and run AI on its own terms, so that the computers, models and know-how it relies on cannot be withheld or dictated by foreign companies or governments.
- open-weight modelopen-weight model, n.: An AI model whose trained internal numbers are published for anyone to download and run on their own computers, even if the data and code used to make it stay private.
Sources
- Bureau of Industry and Security (2025). Framework for Artificial Intelligence Diffusion. Interim final rule, 90 Fed. Reg. 4544 (January 15, 2025).
- Akin Gump (2025). BIS Rescinds Its AI Diffusion Rule and Issues Compliance Guidance Regarding Advanced Computing Items.
Defined in law or standards
Framework for Artificial Intelligence Diffusion, 90 Fed. Reg. 4544 (January 15, 2025); rescission announced by BIS on May 13, 2025
Mentioned in
export controls on AI chipsexport controls on AI chips, n.: Government rules that limit which countries and companies may buy the most advanced AI chips and the machines used to make them. The United States uses them mainly to slow China's progress in AI., compute governancecompute governance, n.: Using control over the computer chips and data centers that AI depends on — who can buy them, where they go, how they are used — as a way to oversee powerful AI., compute thresholdcompute threshold, n.: A legal cutoff based on how much computation was used to train an AI model. Models trained with more than that amount face extra rules, on the theory that bigger training runs produce more capable models., export control directiveexport control directive, n.: A US government order, used against AI models in June 2026, telling a company to stop foreigners from using some of its AI models for national-security reasons., know-your-customer for computeknow-your-customer for compute, n.: Rules that would make cloud companies check who their customers are before renting them enough computing power to train a large AI model, much as banks must verify who opens an account.