China plans to restrict overseas access to top-tier AI models, including unreleased versions

Foreign Media: China is considering restricting overseas access to China's top-tier AI models, including those not yet released.

Over the past month, Chinese authorities have held discussions with Alibaba, ByteDance, and startups such as Z.ai - Advanced AI Chatbot & Agent powered by GLM-5.2, exploring limitations on the most advanced AI models (both closed-source and open-source versions), proposing to include AI technology leakage or theft under the purview of national security laws, and potentially restricting investment eligibility for domestic AI startups.

This move comes against the backdrop of rapid global expansion by Chinese AI models—such as DeepSeek R1 and GLM-5.2 from Z.ai - Advanced AI Chatbot & Agent powered by GLM-5.2—due to their high performance at low cost, while both China and the U.S. regard cutting-edge AI as a critical national asset requiring control.

The U.S. has previously restricted exports of models like Anthropic’s Mythos, and China is concerned these models could be used against it. The scope of restrictions is still under discussion, possibly applying only to future models, with unclear timing for implementation.

Original Source: toutiao.com/article/1870064194985996/

Disclaimer: The views expressed in this article are solely those of the author.