Zhipu founder opposes Beijing’s plan to curb overseas AI access

Chinese authorities are weighing whether to limit foreign access to the country's most advanced artificial intelligence models, a move that would mark a reversal from the open-source strategy that has driven Chinese AI firms' rapid global expansion.

Ministry of Commerce Leads Discussions

China's Ministry of Commerce has held meetings over the past month with executives from Alibaba ↘ 4.74%, ByteDance, and AI startup Zhipu (also known as Z.ai) about potentially restricting overseas access to current and future AI models, Reuters reported on July 7, citing three people familiar with the discussions. No final decisions have been made, and the ministries involved have not commented publicly.reuters+1

Officials discussed making any leak or theft of proprietary AI technology an offence under China's national security law and raised the possibility of new measures restricting who can fund domestic AI startups. The scope of potential restrictions remains under discussion, with two sources telling Reuters the measures may only apply to future models.reuters

A Tiered Restriction Framework

A framework for the proposed controls has already been outlined. Legal scholars participating in a May policy discussion proposed categorizing AI systems by sensitivity: basic open-source tools would require only a simple filing, more advanced models would face security reviews, and the most capable frontier systems could be barred from public release or restricted to domestic use. The proposals were published in an official Supreme People's Court journal.techstrong+1

The discussions come as Chinese open-source models have gained global traction. Zhipu's GLM 5.2, released in mid-June, now performs within a percentage point of Anthropic's top models on key benchmarks while remaining free to download. The model's launch followed the Trump administration's decision in early June to block non-Americans from using Anthropic's most powerful closed models, Mythos 5 and Fable 5.cnbc+2

Zhipu's Founder Pushes Back

Not all Chinese AI leaders support the proposed restrictions. Zhipu founder Jie Tang has publicly advocated for keeping frontier AI openly accessible rather than restricted to a small group of developers. His stance puts him at odds with the security-minded approach being discussed in Beijing, and reflects a broader tension within China's tech sector between those who see open-source AI as a competitive weapon and those who view it as a national security vulnerability.finance.yahoo+1

The deliberations mirror the approach taken by the United States, which has imposed its own restrictions on advanced AI access. George Washington University experts noted that restrictions could affect companies worldwide that rely on China's free AI models as cheaper alternatives to American technology.mediarelations.gwu