- Introduction and sovereign trends to restrict technology
- Ministry of Commerce sponsors confidential meetings and discussions
- Multi-layered restrictive framework under consideration
- Courageous internal resistance and anticipated global warnings
- FAQs
Introduction and sovereign trends to restrict technology
In a move signaling major geopolitical and technological changes, authorities and regulators in China are currently seriously and deeply considering whether to resort to strict restrictions limiting the ability of foreign entities and users to access the country’s most advanced and sophisticated artificial intelligence models and algorithms. If approved and activated, this sovereign step represents a notable retreat and a sharp, historical deviation from the flexible open-source technology strategy previously adopted by the country, which empowered and pushed Chinese AI companies toward rapid global expansion and a prominent, dominant presence on the international technological scene in a short period. This has begun to spark sharp debates and strong opposition from some leaders and thinkers in the Chinese tech sector.
Ministry of Commerce sponsors confidential meetings and discussions
Reports revealed that China’s Ministry of Commerce has already held a series of intensive, closed-door meetings over the past month with senior executives from market-leading tech giants such as Alibaba Group, popular app owner ByteDance, and leading AI startup “Zhipu” to discuss the fine details and potential impacts of restricting and regulating external access to current and future AI models. According to an exclusive report by Reuters on July 7 based on statements from three people closely familiar with these sensitive discussions, no final or official decisions have been made yet, and the concerned ministries and authorities have not publicly commented on this thorny file. During these sessions, regulatory officials discussed the possibility of making any unauthorized leakage or theft of locally owned AI technology and algorithms a serious crime punishable severely under China’s National Security Law, and raised the possibility of implementing new procedures, measures, and legislation that strictly restrict and determine who has the right to fund and inject investments into local AI startups. The scope and impact of potential restrictions remain under discussion and ongoing review, as two sources familiar with the matter told Reuters that harsh measures and bans might apply only to future technological models to avoid causing an immediate shock or disruption in the current global market.
Multi-layered restrictive framework under consideration
The broad outlines of a comprehensive and systematic framework for the proposed controls have already been drawn up. Researchers and legal scholars who participated in an in-depth political discussion last May proposed classifying and passing AI systems according to their strategic and security sensitivity: basic and open-source tools require only a routine filing and a simple permit, while more advanced and capable models will face strict and frequent security reviews and audits. Meanwhile, the leading and most sophisticated systems can be decisively banned from public release abroad or strictly restricted to internal use within China only. These proposals and theories were officially published in a journal affiliated with China’s Supreme People’s Court to pave the legal path for them. These discussions and accelerating political trends come at a time when Chinese open-source models have gained wide and influential global popularity and momentum. Zhipu’s powerful model “GLM-5.2” was released in mid-June, offering astonishing performance that competes with and comes within just one percentage point of the best and most efficient models from American company Anthropic in key benchmarks, while remaining available for free download and open access. The launch of this advanced Chinese model followed a strict decision by the Trump administration in early June banning non-Americans from using Anthropic’s closed and most powerful models, including the Mythos 5 and Fable 5 versions.
Courageous internal resistance and anticipated global warnings
Amidst this atmosphere, not all AI leaders and entrepreneurs in China agree with the proposed restrictions and the new security approach adopted by the authorities. Zhipu founder Ji Tang has publicly and strongly defended the idea and value of keeping leading and advanced AI models available for open and free use by all innovators rather than restricting and limiting them to a small, closed group of developers. This dissenting and explicit stance puts Tang in direct and heated confrontation with the national security-focused approach being firmly discussed in Beijing, and also reflects a deeper tension and intellectual and technological struggle within the Chinese tech sector itself between those who view open-source AI tools as an effective competitive weapon and soft power to conquer markets, and those who view them as a serious weakness and loophole that could threaten national security. These Chinese deliberations clearly reflect and align with a similar approach taken by the United States, which has already imposed strict restrictions on its adversaries’ access to advanced AI technologies. Experts and analysts at George Washington University pointed out that these potential restrictions from both sides could negatively impact and disrupt the trajectory of companies worldwide, especially startups currently relying on free Chinese AI models as powerful and cheaper alternatives to expensive and closed American technology.
FAQs
Question: What measures is the Chinese government studying regarding smart models?
Answer: The government is studying the implementation of new policies to restrict and limit foreign users’ and entities’ access to the latest and most powerful advanced AI models designed in China.
Question: Why does the founder of Zhipu oppose these government measures?
Answer: He opposes them out of his firm belief that models should remain open-source and globally available to foster an innovation environment and support technical progress rather than being monopolized.
Question: How will future models be handled according to the leaked plans?
Answer: It is proposed to divide them into several categories where leading and future models will be subject to strict security scrutiny and may be banned from export or public release outside the country.
