Chinese authorities have held meetings with leading tech firms including Alibaba, ByteDance, and Z.ai to discuss potentially restricting overseas access to China’s most advanced AI models. These talks, led by the Ministry of Commerce, follow prior steps to keep homegrown AI domestic and treat cutting-edge AI as a critical national asset requiring controls. Discussions covered limits on both closed-source and open-weight models, making leaks or theft of proprietary AI an offense under national security law, and possible restrictions on funding for domestic AI startups.
The scope remains under discussion and may apply only to future models, with no clear timeline for implementation. Since DeepSeek’s R1, Chinese models have gained global traction due to low costs and rising capabilities; restrictions could raise costs for international businesses. Alibaba’s Qwen, ByteDance’s Doubao, and Z.ai’s GLM-5.2 are among the prominent models affected. Officials are particularly concerned about tools like Mythos being used against Chinese interests.
This mirrors U.S. actions, such as limiting access to Anthropic’s advanced models. China has already taken measures including ordering Meta to unwind its Manus acquisition, issuing rules on overseas deals involving Chinese tech and data, and investigating startups that moved abroad for potential export control violations. Legal experts have proposed a tiered system for open-source AI, from simple filing for basic tools to bans or domestic-only restrictions for frontier models.
These developments underscore that major nations now view AI advances as strategically vital rather than mere hype, comparable in impact to the harnessing of electricity, with the world still at the early stages of this technological shift.