Publication date: 16 October 2026
China has proposed adding 40 technologies in artificial intelligence, biomedicine and solar manufacturing to its catalogue of technologies prohibited or restricted from export. The most consequential part for the technology industry is the inclusion of 23 AI-related areas covering large models, agents, computing clusters and information processing.
The proposal, issued jointly by the commerce and science ministries on 16 October, is open for public comment until 15 November. But it is not merely a future consultation. During the interim period, exporters must register covered transfers before they take place, including transfers of intellectual-property ownership, licences for overseas use and changes that place a domestic technology owner or controller under a foreign entity.
The immediate effect is therefore procedural visibility rather than a blanket ban. Registration gives the government a clearer view of who is transferring sensitive technology, to whom and for what purpose. The eventual catalogue could then distinguish prohibited technologies from those requiring licences.
From controlling hardware to controlling capability
Most international attention on AI trade has focused on advanced chips and the machinery used to manufacture them. China’s proposal widens the field. Models, agent systems, cluster designs and specialist information-processing methods can cross borders through licences, cloud operations, partnerships and corporate transactions even when no physical product is shipped.
This makes the policy significant for Chinese technology companies expanding abroad. A company may be able to sell access to a service without transferring its underlying technology, while a joint venture, overseas deployment or intellectual-property licence could trigger review. The precise boundary will matter for deal structures, research partnerships and the design of international cloud operations.
The ministries have also created important exclusions and fast paths. Technology already placed in the public domain through open source or academic publication is outside the proposed controls. The commerce ministry said automatic-licensing or other facilitated treatment would apply to some open-source activity, Chinese-owned overseas computing centres, overseas chip fabrication, collaborative research and international standards work.
Openness becomes a regulatory category
Those exceptions reveal the policy’s deeper tension. China wants to retain strategic control over advanced capability without isolating its developers from the international research and commercial networks that help technology improve. Open publication is not simply a technical choice under this framework; it may affect whether a transfer falls inside the control system.
Companies will need to map which parts of their AI stacks are public, proprietary or subject to third-party rights. They will also need governance that connects engineering decisions to export compliance. A model release, overseas fine-tuning project or change in corporate control can no longer be treated as a purely product or investment decision if restricted know-how may move across borders.
The broader geopolitical implication is that technology controls are becoming reciprocal and more intangible. Washington’s restrictions have centred heavily on computing hardware sold to China. Beijing is now signalling that domestic AI know-how can also be treated as a strategic asset. That does not make the two systems identical, but it does make global AI operations more dependent on two sets of national-security rules.
For leaders, the practical response is not to assume that all cross-border AI collaboration will stop. It is to recognise that architecture, licensing and documentation now determine strategic exposure. The next phase of the technology contest will be fought not only over who can buy the best chips, but over who may transfer the knowledge that turns computing power into capability.