China has proposed adding AI models, intelligent agents and computing clusters to its catalogue of technologies that are prohibited or restricted from export. Why it matters: if adopted, some of the core building blocks of China’s AI industry could face export controls and pre-export registration, at a time of heightened competition with the United States over frontier technology.
October 16, 2026
What happened?
China’s commerce ministry announced on Friday that it is proposing to add 40 technologies to its export control catalogue. The technologies span three sectors: artificial intelligence, biomedicine and photovoltaics. Inclusion in the catalogue would mean a technology is either prohibited from export or restricted, rather than freely transferable overseas.
This is a proposal, not a final rule. The ministry is seeking public opinions on the revision until November 15.
Which AI technologies are named?
The AI items named in the proposal are broad rather than narrow. They include AI models themselves, intelligent agents, and computing clusters. Reuters described those three categories as fundamental building blocks of China’s AI industry, which underlines why the proposal matters beyond any single company or product: it reaches the models, the agent software built on them, and the clustered computing used to train and run them.
Alongside the consultation, China is putting interim pre-export registration management in place for the relevant technologies. In practical terms, that means affected exports may need to be registered before they ship while the revision is still being considered, rather than waiting for a final catalogue decision.
What did the ministry say about why?
In a separate statement, the ministry linked the proposal to the expanding use of the related technologies in sensitive fields. It characterised the revision as a routine update and said it does not target any specific country, region or entity.
That framing matters, but so does the context Reuters noted: the proposal lands amid increasing competition and geopolitical tension with the United States over frontier technology development. Export controls on models and compute have become one of the main levers governments use in that contest, on both sides.
Why it matters
For AI companies and researchers, the key point is scope. Controls aimed at finished products are one thing; controls that reach models, agents and computing clusters could affect licensing deals, cross-border deployments, partnerships and how Chinese AI technology is offered outside China. The details will decide the real impact — which specific technologies end up listed, whether they are prohibited or merely restricted, and what the registration process requires in practice.
For everyone watching the US-China AI race, this is another sign that the contest is moving further into formal export rules, not just tariffs and chip shipments. Tested Intelligence will follow what happens after the November 15 comment deadline — that is when a proposal like this either becomes policy, gets narrowed, or stalls.
Frequently asked questions
Is this a ban on exporting Chinese AI models?
Not yet, and not necessarily. The ministry has proposed adding the technologies to a catalogue of items that are prohibited or restricted from export. Until the final revision is published, it is not clear which items would be prohibited outright and which would be restricted or subject to registration.
When could this take effect?
Public opinions are being sought until November 15. Separately, interim pre-export registration management for the relevant technologies is being implemented now, according to the ministry announcement reported by Reuters.
Does the proposal target the United States?
The ministry said the revision does not target any specific country, region or entity and described it as routine. Reuters reported it in the context of increasing US-China competition over frontier technology.
Sources: Reuters