In recent weeks, China's Ministry of Commerce has held a series of closed consultations with leading tech giants — Alibaba, ByteDance, and Z.ai. The focus of discussion was a potential package of restrictions aimed at blocking foreign access to China's most advanced AI models. This concerns future developments — solutions already released to the market, apparently, will not be subject to these measures.
Authorities are considering classifying the leakage or theft of proprietary AI technologies as a violation of national security law. Separately, the introduction of restrictions for foreign investors wishing to invest in Chinese AI startups is being discussed. This is a logical step within the strategy of technological sovereignty, but it also stems from a specific concern: Beijing fears that American cyber models, such as Mythos, could be used to find and exploit vulnerabilities in Chinese infrastructure in Washington's interests.
Such decisions carry serious risks for China's global competitiveness. It was precisely open access, low cost, and a less stringent distribution regime in 2025–2026 that allowed Chinese models to actively capture the international market. Restrictions could undermine this key advantage, weakening one of the main channels through which China counters American developments.
Context of the tightening: from the Meta-Manus deal to Claude distillation
This is not the first time China has demonstrated its determination to control the AI sector. In April, the National Development and Reform Commission ordered the cancellation of Meta's deal to acquire the startup Manus for $2 billion. In June, Meta stopped data exchange and disconnected Manus from its systems, and banned employees from using its tools. That same month, Beijing tightened control over foreign transactions involving Chinese investors, technologies, and data.
These events occur against the backdrop of the expansion of U.S. export controls from chips to access to AI models themselves. On June 12, Anthropic, at the request of U.S. authorities, restricted access to Fable 5 and Mythos 5 for foreign nationals. On June 30, the directive was lifted, but before that, the startup accused operators linked to Alibaba of large-scale distillation of Claude — the largest known attack of its kind, aimed at accelerating the development of Chinese models.
At the end of June, an independent developer discovered mechanisms in Claude Code that read the local time zone and checked addresses for ties to Chinese cloud providers. After that, Alibaba banned employees from using Anthropic products in work environments starting July 10, suggesting they switch to its own Qoder platform.
In parallel, China is discussing a differentiated approach to open source: from simple registration for basic tools to a complete ban on public release or restriction to the domestic market for the most sensitive models. These discussions are taking place at the Intellectual Property Court of the Supreme People's Court of China.
My analysis: We are witnessing the formation of two isolated AI ecosystems — Chinese and Western. Beijing's restrictions may accelerate this process, but in the long term, openness was the main trump card of Chinese models. By closing itself off, China risks losing not only the global market but also access to the international research environment, which could slow down its own pace of innovation.