China Considers Restricting Foreign Access to Advanced AI Models, Sources Say

Chinese authorities are exploring measures to restrict overseas access to the nation's most advanced artificial intelligence models, according to three people familiar with recent discussions. The move signals Beijing's growing view of cutting-edge AI as a strategic national asset that requires protection, mirroring similar national security concerns in the United States.

High-Level Discussions on AI Controls

Over the past month, officials from China’s Ministry of Commerce and the National Development and Reform Commission met with representatives from leading tech companies, the sources said. Participants reportedly included tech giants Alibaba and ByteDance, as well as the AI startup Z.ai.

Discussions focused on several potential restrictions:

• Placing limits on the most advanced AI models, including both proprietary and open-weight versions.
• Making the theft or leak of proprietary AI technology an offense under China's strict national security law.
• Implementing new measures to control who can fund domestic AI startups.

The scope and timing of any potential rules remain under discussion, and they may only apply to future models, two of the sources noted. The involved government agencies and companies did not respond to requests for comment from Reuters, which first reported the news.

Market Impact and Global Competition

Ad

Any decision by Beijing to curb access could have significant ripple effects across the global AI market. Chinese models, such as Alibaba's Qwen, ByteDance's Doubao, and Z.ai's highly capable GLM-5.2, have gained international traction due to their increasing sophistication and lower costs compared to Western counterparts. Restricting their availability could increase operating costs for international businesses that have integrated them into their workflows.

The potential controls are being considered as both China and the U.S. escalate efforts to protect their domestic AI ecosystems. The U.S. government recently imposed access restrictions on Anthropic's advanced "Mythos" model over concerns it could be misused by foreign adversaries. According to two sources, Chinese officials are worried that such powerful AI tools could be deployed to exploit software vulnerabilities and harm Chinese interests.

A Pattern of Tightening Control

These talks are part of a broader trend of Beijing tightening its grip on homegrown technology. Earlier this year, China's state planner ordered Meta to unwind its acquisition of the Chinese-founded AI startup Manus. Authorities also issued sweeping new rules in June to control overseas deals involving Chinese technology and data, and have reportedly launched investigations into local AI startups that have moved abroad to determine if they have violated export control laws.

While the exact mechanism for AI export controls is unclear, a potential framework was proposed at a May roundtable of Chinese legal experts. According to a summary published in an official court journal, they suggested a tiered system: simple registration for basic open-source tools, security reviews for more advanced technologies, and a ban on public release or restriction to domestic use for the most sensitive frontier models.