In recent weeks, China's Ministry of Commerce has held a series of consultations with tech giants, including Alibaba, ByteDance, and Z.ai, discussing the introduction of strict restrictions on foreign users' access to the most advanced Chinese artificial intelligence models. This signals that Beijing is ready to shift from expansion to isolation in a strategically important sector.
According to my information, the measures under discussion will only apply to future models, leaving already released products outside the scope of the ban. Officials propose classifying the leakage or theft of proprietary AI technologies as a violation of national security law. Separately, restrictions on foreign investors funding Chinese AI startups are being considered.
Notably, these steps are linked not only to the logic of technological sovereignty but also to growing concerns about American cyber models. Beijing fears that systems such as Anthropic's Mythos could be used to find and exploit vulnerabilities in Washington's interests. This is a mirror response to recent U.S. actions, which have already restricted foreign citizens' access to their models.
The Price of Isolation: Losing the Global Market
Such decisions threaten to weaken one of the main channels through which China competes with American developments. In 2025–2026, Chinese models actively gained global recognition due to low costs, open weights, and a less restrictive access regime. Introducing restrictions could destroy this key advantage, turning China into a closed ecosystem.
Beijing has previously taken individual steps to tighten control. In April, the National Development and Reform Commission canceled Meta's deal to acquire startup Manus for $2 billion. In June, oversight of foreign deals involving Chinese investors was tightened. In May, the Supreme People's Court discussed a differentiated approach to open source—from simple registration for basic tools to a complete ban on public releases for the most sensitive models.
These events unfold against the backdrop of escalating technological confrontation between the U.S. and China. Anthropic has already restricted access to Fable 5 and Mythos 5, then lifted the restrictions. In response, Alibaba banned its employees from using Anthropic products, urging a switch to its own Qoder platform. Meanwhile, 360 Security Technology founder Zhou Hongyi introduced the Tulong Feng tool as a "Chinese response" to Mythos 5.
Expert commentary: We are witnessing the formation of two parallel AI universes—Chinese and American. Restricting access to models is just the first step. In the coming years, we will see complete market fragmentation, where each side builds its own ecosystem, and global projects face insurmountable regulatory barriers. For investors, this means that betting on a "single global AI" becomes extremely risky.