know-your-customer for compute
In plain wordsRules 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.
A regime requiring providers of large-scale computing — chiefly cloud or infrastructure-as-a-service (IaaS) firms — to verify customers' identities, assess their purposes and keep records, and in some versions to report customers whose usage indicates training of a frontier-scale model. Modeled on anti-money-laundering KYC duties in banking, it targets the point where compute is rented rather than bought, which chip export controlsexport 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. alone do not reach: a restricted party can train a model in a foreign data center without importing a single chip. Executive Order 14110Executive Order 14110, prop. n.: A sweeping 2023 order by US President Joe Biden that told federal agencies how to manage AI risks and required makers of the largest AI models to report their safety tests to the government. It was revoked in 2025. directed the US Commerce Department to require US IaaS providers to verify foreign customers and report foreign training of large models, and a rule was proposed in January 2024; the AI Diffusion RuleAI Diffusion Rule, prop. n.: A 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sorted the world's countries into three groups with different limits on how many advanced AI chips they could buy., rescinded before taking effect, would have attached related security and customer-screening conditions to licensed data centers abroad. Concerns include customer privacy, compliance cost and evasion through resellers and shell companies.
Under compute KYC, a cloud provider would flag a new customer reserving ten thousand GPUs for three months.
Origin
Proposed for frontier AI by Janet Egan and Lennart Heim in 'Oversight for Frontier AI through a Know-Your-Customer Scheme for Compute Providers' (2023), building on Executive Order 13984 (January 2021), which sought identity checks by US IaaS providers to counter malicious cyber actors.
Usage note
The January 2024 proposal had not been finalized when the administration changed in January 2025; since then the idea has surfaced mainly in export-licensing conditions and legislative proposals rather than in a general rule.
Related
- 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.
- 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.
- on-chip governance mechanismson-chip governance mechanisms, n.: Features built into AI chips that could check or enforce rules about how they are used — for example, proving which country a chip is in, or refusing to run without a valid license.
- AI Diffusion RuleAI Diffusion Rule, prop. n.: A 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.
- 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.
Sources
- Egan, J. & Heim, L. (2023). Oversight for Frontier AI through a Know-Your-Customer Scheme for Compute Providers. arXiv:2310.13625.
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., Executive Order 14110Executive Order 14110, prop. n.: A sweeping 2023 order by US President Joe Biden that told federal agencies how to manage AI risks and required makers of the largest AI models to report their safety tests to the government. It was revoked in 2025., AI Diffusion RuleAI Diffusion Rule, prop. n.: A 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., on-chip governance mechanismson-chip governance mechanisms, n.: Features built into AI chips that could check or enforce rules about how they are used — for example, proving which country a chip is in, or refusing to run without a valid license., structured accessstructured access, n.: Letting people use an AI system through a controlled channel, such as a website or programming interface, instead of handing over the model itself, so that the developer can limit misuse.