China Weighs 'Silicon Curtain' on Its Top AI Models

Chinese authorities have held closed-door meetings with Alibaba, ByteDance, and Z.ai about potentially restricting overseas access to the country's most advanced AI models. Officials discussed making AI theft punishable under national security law, mirroring US export controls on American AI.

Key Takeaways

Chinese authorities have held closed-door meetings with Alibaba, ByteDance, and Z.ai about potentially restricting overseas access to the country's most advanced AI models.
Officials discussed making AI theft punishable under national security law, mirroring US export controls on American AI.

AI synthesis2 sources19

Chinese authorities have held meetings over the past month with top tech firms including Alibaba, ByteDance, and rising startup Z.ai, discussing potential restrictions on overseas access to China's most advanced AI models — a move that would effectively erect a 'silicon curtain' around the country's artificial intelligence crown jewels.

According to three people familiar with the discussions, the talks — led by China's Ministry of Commerce — explored putting limits on both closed-source and open-weight frontier models, including those yet to be released. Officials raised the possibility of making any leak or theft of proprietary AI technology an offense under China's stringent national security law, one source said, and floated new measures to restrict who can fund domestic AI startups.

The scope of potential restrictions is still being discussed and may only apply to future models. But the implications would ripple across global AI markets: Alibaba's Qwen and ByteDance's Doubao are among the most widely used AI models globally, and Z.ai's GLM-5.2 has recently been closing the gap with leading U.S. offerings at a fraction of the cost.

The talks mirror similar actions by the Trump administration, which in June ordered that foreign nationals be barred from accessing Anthropic's most advanced Fable and Mythos models. Export controls for Fable have since been lifted after new safeguards were put in place, but Mythos — designed for cybersecurity professionals — remains restricted to 'trusted' U.S. organizations.

Chinese authorities are deeply worried about the potential for Mythos to exploit software vulnerabilities against Chinese interests, according to two sources. Cybersecurity firm 360's founder Zhou Hongyi has publicly called for China to develop its own equivalent.

The discussions follow a series of steps by Beijing to keep homegrown AI within its borders, including ordering Meta in April to unwind its $2 billion acquisition of Chinese-founded AI startup Manus, and issuing sweeping new rules in June tightening control over overseas deals involving Chinese investors and technology.