China Proposes New Export Controls On 23 Advanced AI Technologies

China proposes export controls on 23 AI technologies

China has proposed adding export controls covering 23 artificial-intelligence technologies that involve models, agents and computing clusters, according to a Reuters report. The proposed measures are part of a broader revision to the country’s export-control catalogue that was disclosed by the commerce ministry and opened for public comment.

China AI export controls October 2026 was cited as a key move in the proposal, which aims to address perceived misuse and national-security risks associated with advanced AI technologies. Reuters reported the proposals but did not indicate that the measures are final; the ministry said the changes are subject to a public consultation that closes on November 15.

The ministry described the rationale for the proposal in its public materials, attributing it to concerns about misuse and national-security risks. Reuters is the source for the reported details of the proposal, which specifically lists 23 AI technologies among 40 items under consideration across AI, biotechnology and solar power.

What the proposal would cover

According to the Reuters report, the draft export controls would target overseas deployment of models that contain restricted technology, and would explicitly include agents and computing clusters. The proposals were said to cover the physical and service-related aspects of deploying certain advanced AI capabilities beyond China’s borders.

Importantly, the draft rules would exclude several common access methods: API access, websites, mobile apps and software-as-a-service (SaaS) offerings would not fall under the proposed overseas-deployment restrictions for the listed technologies, Reuters reported. That carve-out signals a narrower focus on deployments that likely involve transfer of controlled components or capabilities, rather than routine remote access to services.

Context within a wider catalogue revision

The 23 AI items are part of a larger proposed revision to China’s export-control catalogue covering 40 technologies across three sectors: artificial intelligence, biotechnology and solar power. Reuters reported the catalogue update as the vehicle through which the commerce ministry is seeking to regulate more advanced or sensitive technologies.

Public consultation on the proposed catalogue changes is scheduled to close on November 15, according to the commerce ministry’s published timeline cited by Reuters. The outcome of the consultation and any subsequent changes to the catalogue were not detailed in the report.

Key implications highlighted by the report

Targeted controls: The proposal focuses on specific AI artefacts—models, agents and computing clusters—rather than broad bans on remote access services, given the stated API, website, app and SaaS exclusions.
Security rationale: The commerce ministry framed the changes as necessary to guard against misuse and national-security risks, as reported by Reuters.
Consultation period: Stakeholders have until November 15 to submit comments, after which the ministry may revise the proposal before making it final or implementing it.

What remains unclear from the report

The Reuters report provides the proposed scope and timing of the consultation but does not specify the enforcement mechanisms, penalties for non-compliance, or the technical criteria that will define which models, agents or clusters are deemed restricted. The report also does not offer detail on how service providers or foreign companies operating in China might be required to comply if the proposals are adopted.

Further details would be necessary to assess how the measures might affect cross-border AI research collaborations, cloud-service deployments, and multinational AI firms that interact with Chinese entities. Reuters’ coverage conveys the ministry’s stated concerns but leaves open many operational and legal questions that a finalized measure would need to address.

How this fits into broader technology policy moves

The proposed controls form part of a wider pattern of many governments updating export-control regimes to address emerging technologies. Reuters reported that the proposed catalogue revision includes items across AI, biotechnology and solar power, indicating a broader assessment by Chinese authorities of technologies they consider sensitive or strategically important.

The proposals, as reported, appear aimed at limiting certain transfers of advanced capabilities that the ministry judges could be misused or pose national-security risks. How those judgements will be operationalised—and how other governments and companies will respond—remains to be seen pending the consultation outcome and any subsequent regulatory steps.

Short FAQ

Who reported the proposal? Reuters reported the details of the proposed export controls and the consultation timeline.
What technologies are covered? The proposal would add 23 AI technologies to an export-control catalogue, specifically referencing models, agents and computing clusters, per Reuters.
Are APIs or SaaS covered? Reuters reported the draft would exclude API, website, app and software-as-a-service access from the overseas deployment restrictions.
When does public consultation end? The commerce ministry’s public consultation period closes on November 15, according to Reuters.

Related References

Reporting in this article is based on a Reuters report and information released by China’s commerce ministry as cited therein.

What China’s proposed AI export controls would cover

The proposal is important because it treats advanced AI as a strategic technology rather than only a commercial software product. Models, agents and computing clusters can involve training data, model weights, specialized hardware and large-scale infrastructure. Controls on overseas deployment could affect how Chinese developers license or operate certain systems outside the country.

Reuters reported that the proposal would exclude access through APIs, websites, apps and software-as-a-service platforms. That distinction means a company may still be able to offer some AI functionality internationally even if transferring a restricted model or its underlying technology requires approval. The final rules will determine how officials define sensitive technology and which transactions need registration or a license.

Why the proposal matters for businesses

AI developers, cloud providers and multinational companies will need to watch the consultation process closely. The effects could include additional compliance checks, slower technology transfers, different partnership structures and more pressure to keep model development inside approved jurisdictions. Open-source technology and academic material may receive different treatment, but companies should not assume that every public model is automatically outside the rules.

The proposal also fits into a wider U.S.-China competition over chips, models and computing capacity. It is not the same as a final export ban. Public comments are scheduled to close on November 15, after which the catalogue and implementation details may change.