Yuyuan Tantian interprets China's new AI technology export rules: what is controlled, what is not

Yuyuan Tantian | Tan Zhu | Editor: Wang Tianyong | 2026-10-23 07:56

① On October 16, the Ministry of Commerce and the Ministry of Science and Technology publicly solicited opinions on the revision of the Catalogue of Technologies Prohibited or Restricted from Export from China, which includes 23 artificial intelligence technologies; ② When enterprises use restricted large models to provide external API calls, APP/webpage interaction, SaaS applications and other such services, none of these fall within the scope of large-model technology export management.

This question can be traced back to October 16. On that day, the Ministry of Commerce and the Ministry of Science and Technology publicly solicited opinions on the revision of the Catalogue of Technologies Prohibited or Restricted from Export from China (hereinafter referred to as the Catalogue). Among them are 23 artificial intelligence technologies, including large models, agents, computing power clusters, information processing and others.

Tan Zhu has learned that since the policy was announced, two leading enterprises, DeepSeek and MiniMax, have smoothly completed registration procedures and completed open-source and overseas authorized use.

In the first half of this year, relevant departments began brewing the draft for soliciting opinions on this Catalogue, and beforehand widely organized research institutions, industry organizations and leading enterprises in the industry to carry out in-depth assessments.

Lawyers Liu Jing and Liu Hongmei of Beijing Dacheng Law Offices participated in these discussions. They shared a detail with Tan Zhu: compared with the initial draft from several months ago, the control points of some technology items in the AI field kept being adjusted in line with the latest industry developments. Many of the concerns and suggestions raised by enterprises at the meetings were well reflected in the policy released this time.

What they care about more now is whether there will be clearer operational guidelines going forward.

Another detail is that in the relevant Q&A, the Ministry of Commerce's explanation of "what is not regulated" is noticeably more extensive and specific than its explanation of "what is regulated." This also shows that the core of the Catalogue is to draw boundaries, rather than simply restrict the export of AI technology.

Tan Zhu has also sorted out several key points to help everyone understand how this set of rules works.

This set of rules

what exactly does it regulate?

People who do not work in technology can easily confuse "technology export" with "product export." Some may ask: does using restricted large models to provide external API calls, APP/webpage interaction, SaaS applications and the like fall within the scope of large-model technology export management?

Large models that provide services after being trained on China's core data are proposed to be prohibited from export; large models that provide services after being trained on important data are proposed to be restricted from export. Simply put, as long as the model training process involves concerns about national data security, then regardless of whether the model itself is large or small or how many parameters it has, it will be subject to export constraints.

Second, sensitive uses.

For example, agent technology that reaches a certain level of performance is itself proposed to be restricted from export; but if this technology is specifically or mainly used in highly sensitive scenarios such as national critical information infrastructure, national defense and military industry, and core government affairs, then its nature changes and it is directly upgraded to proposed prohibition of export. The logic conveyed here is very clear: where the same technology is used and for whom it serves determines how strict the regulation it faces will be.

Third, the computing power foundation.

After reading these three points, one can understand that this set of rules regulates the underlying logic of technology.

When enterprises use restricted large models to provide external API calls, APP/webpage interaction, SaaS applications, lightweight deployment on local devices and other such services, none of these fall within the scope of large-model technology export management.

Taking into account the practical needs of international cooperation, the Catalogue also sets up two types of facilitation arrangements.

One type is open-source technology and specific international business. For technology exports in situations such as open-sourcing artificial intelligence technologies, Chinese enterprises building and operating their own computing power centers overseas, going abroad to carry out chip tape-out manufacturing, using overseas resources for joint research and development, and participating in the formulation of international standards, facilitation measures such as automatic licensing apply, basically achieving "report and export immediately"; for relevant technologies that have already entered the public domain through open source, publication of academic papers and other means, it is proposed not to include them in the scope of export management.

The other type is international artificial intelligence cooperation. For technology exports needed by members of the World Artificial Intelligence Cooperation Organization to carry out international cooperation on capacity building, technology sharing and joint rule-building, more facilitation and support will be provided.

Liu Hongmei told Tan Zhu that after the new rules take effect, technologies in the prohibited export category must not be exported; for technologies in the restricted export category, enterprises can submit applications to the competent authorities and strive to obtain licenses, and this will not disrupt enterprises' original overseas layout and business plans.

This also confirms the domestic position of "encouraging openness." In the face of the new rules, enterprises can still remain composed.

Why push out this set of rules now?

First, differentiated treatment by country. In January 2025, the U.S. Department of Commerce's "Framework for Artificial Intelligence Diffusion" had designed management of advanced computing chips and some advanced model weights based on the "sensitivity" of destination regions and other factors, strictly controlling the flow of advanced AI capabilities to specific entities.

Second, dividing circles by camp. The CEO of Anthropic once proposed a kind of "alliance strategy": uniting a group of "democratic countries" to form a small circle, controlling the AI technology supply chain, drawing more countries in, and isolating opponents.

But China is not taking this path.

This revision is routine work carried out in accordance with laws and regulations. Since 2001, China has adjusted and revised the Catalogue many times.

Tan Zhu has learned that the focus of management is the act of technology export, not the technology itself.

This approach requires policy to break down control standards in sufficient detail and turn them into objective, quantifiable provisions.

Why can China do this?

Since the standards must be detailed and objective, the matter of "risk assessment" must be made independently verifiable. This leads to a deeper proposition:

Who has the capacity to assess risk and implement governance?

On this question, the premise presupposed by the United States is that security can be guaranteed only if technological capabilities are held in the hands of a few.

But from the perspective of technological development, leaders do indeed have an advantage when training frontier large models, yet the development of open source is shaking this premise at its root.

First, the threshold for training large models is very high, but the threshold for conducting security testing is not that high.

According to industry estimates, training top-tier large models often requires investing tens of thousands of computing cards, but to assess their security risks, there is absolutely no need to spend heavily on retraining from scratch. This breaks the cognitive limitation that "only technological leaders are qualified to conduct risk reviews." More research institutions and even government departments now have the conditions to conduct security testing on large models and reproduce security issues.

Second, relying on open source, governance capabilities are rapidly becoming widespread.

After weights are opened, models are no longer entirely "black boxes." Security researchers and small and medium-sized technical teams around the world can polish security tools at relatively low cost.

Since technical-level problems can be solved, rule design can pursue more precise implementation results.

The purpose of introducing this temporary management measure of pre-export registration is to grasp in advance the demand for relevant technology exports and make regulation more scientific.

Moreover, to give enterprises a clear and unified set of judgment standards, the Catalogue also sets explicit quantitative indicators: for example, computing power exceeding 10²⁶ floating-point operations, parameters exceeding 2 trillion, and capabilities reaching the global top ten level.

In this way, enterprises no longer need to guess the direction of regulation, and the government can direct limited administrative resources to the links where risks truly exist.

Why build a technology import and export management system?

In the past, when facing external technology blockades, we were more on the defensive. But as China moves to the forefront of more and more technological fields, how to manage technology has become a new issue.

In this role shift, how to avoid affecting normal commercial cooperation is a problem that had not been fully solved before.

Therefore, the focus of this policy adjustment is not on "how many technologies have been controlled," but on building technology import and export management into a national capability that can operate with precision.

In the future, it will support our participation in the global technology game at two levels:

At the micro level, achieving targeted interception.

Undercurrents still exist in the game in the international technology field, and the pressure to prevent core technologies from being illegally transferred and to curb sensitive technologies from being disguised and leaked through commercial cooperation has always existed. This management mechanism can enable regulators, within the intricate global commercial network, to accurately identify and cut off technology transfers that may endanger national security.

At the macro level, forming the ability to shape rules.

An increasingly obvious trend is that China is no longer a passive acceptor of international rules. By establishing transparent, non-discriminatory management standards based on objective technical indicators, China's rules can provide a reference for international technology governance.

In this process, we can use multilateral platforms to discuss the boundaries of AI governance with all parties. When this set of rules is gradually recognized or aligned by more multilateral cooperation mechanisms, China will have more institutional discourse power in the global technology game.

This technology catalogue is only a small step for China in filling the shortcomings of technology import and export management. But the direction it points to is clear: within the same institutional framework, security bottom lines and open cooperation can absolutely coexist.

Artificial intelligence

Cailianshe statement: The content of the article is for reference only and does not constitute investment advice. Investors operate at their own risk based on this.

Business cooperation

Popular unlocks

[Master ETFs] Leading companies are steadily advancing the pace of mass production; the industry is currently at the dawn moment of moving from technological breakthroughs to large-scale commercialization, and with policy and capital support, future market space is broad

① Leading companies are steadily advancing the pace of mass production; the industry is currently at the dawn moment of moving from technological breakthroughs to large-scale commercialization, and with policy and capital support, future market space is broad; ② Institutions: The third-quarter report is an important verification window, mainly observing these two directions.

[Nine O'Clock Special] DeepSeek plans to raise about 100 billion yuan in the latest financing round; analysts are optimistic that large models will accelerate breakthroughs in embodied "brain" capabilities and the implementation of general embodied intelligence; PHP has a serious security vulnerability, and websites of institutions in multiple countries have been temporarily taken offline

① DeepSeek plans to raise about 100 billion yuan in the latest financing round; analysts are optimistic that large models will accelerate breakthroughs in embodied "brain" capabilities and the implementation of general embodied intelligence. This company already has a small number of orders for robot structural parts; ② Websites of institutions in multiple countries have been temporarily taken offline! PHP has a serious security vulnerability. Analysts believe that in the AI era, intensified threats and the continuous expansion of the application scope of AI technology are expected to become the main driving forces. This company's AI security order amount in the telecom operator industry in the first half of the year increased by more than 300% year on year; ③ Trump temporarily suspended additional tariffs on wind turbines, and the 180-day consultation period leaves room for follow-up.

[Touch Gold Interactive Easy] Humanoid robots + consumer electronics, four domestic locations and overseas bases have officially been put into production, mass production and delivery of humanoid robot parts and complete machines; this company's consumer electronics products are in the peak delivery season

① Humanoid robots + consumer electronics, four domestic locations and overseas bases have officially been put into production, mass production and delivery of humanoid robot parts and complete machines; this company's consumer electronics products are in the peak delivery season; ② GLP-1 weight loss + innovative drugs, the phase clinical trial of a self-developed Class 1 innovative peptide drug has obtained top-line analysis data, showing positive efficacy and good safety in overweight/obese populations and achieving the expected goal. This company's two Phase III studies for type 2 diabetes are being actively advanced.

Related reading

Technology stocks adjust, is the high-growth logic still there? STAR Market leading companies speak out: full orders, upward prosperity

STAR Market Daily reporter Zeng Le 2026-10-23 08:34

① Both Source Photonics and Lianxun Instruments stated that the company's production and operations, order delivery, and capacity construction are all operating normally. ② With the explosive demand for AI computing power and high-performance computing, GAA, 3D NAND, and 3D DRAM are accelerating in adoption, the complexity of etching processes and the amount of single-wafer equipment used are rising simultaneously, and the global etching equipment market space continues to expand.

Will the memory market go from "chip shortage" to "red ocean"? Jensen Huang speaks out: oversupply in 2029 is unlikely

Cailian Press Bian Chun 2026-10-23 08:06

① Nvidia CEO Jensen Huang said on Thursday that an oversupply of memory by 2029 is unlikely; ② He said, "The great thing about memory is that you can always use more"; HBM will continue to be used in the coming years.

Barclays announces 5,000 layoffs; UK union accuses it of using AI as an excuse to abandon employees

Cailian Press Li Ying 2026-10-23 00:38

① Barclays Bank announced it will cut 5,000 positions, about 5% of its total workforce, with management saying the adjustment is related to organizational streamlining and the impact of AI; ② Barclays will invest about £600 million in this round of restructuring this year, roughly twice its previous estimate, to reduce future costs; ③ UK union Unite criticized Barclays for using AI as an excuse to "abandon employees" and said it is in urgent negotiations with the bank.