China May Restrict AI Exports, But Open-Source Tech Is Safe
New rules could affect the AI tools you use daily—here's what you need to know.
China’s Ministry of Commerce and Ministry of Science and Technology released a draft revision to the Catalog of Technologies Prohibited and Restricted from Export on October 16, 2026, proposing to add 40 technologies across AI, biomedicine, and solar PV—23 of them AI-related, covering foundation/large models, intelligent agents, computing clusters, and information processing. Public comments are open until November 15, and temporary pre-export registration requirements for covered technologies take effect immediately.
According to the article, technologies already made public via open-source releases or academic publications are explicitly exempted and not subject to the controls; open-sourcing AI technologies, plus certain activities such as Chinese firms building their own overseas compute centers, are eligible for streamlined automatic licensing; providing access via APIs, apps, websites, or SaaS is generally not treated as a technology export; and overseas local deployment of restricted models can fall under the rules, with possible facilitation in lower-risk cases such as a company’s own overseas cloud. This is not a blanket ban on open-weight models, but new models not yet released may face registration or approval steps before their weights or source code can be transferred abroad, with thresholds based on performance, parameters, data sensitivity, and similar factors still being defined. Chinese labs such as DeepSeek and Qwen could see additional compliance requirements on future open releases.
- Open-source AI is exempt, so free AI tools you use today won't be restricted.
- Future AI models from China may need government approval before being shared globally.
- Using AI through apps or websites is not considered an export, so services continue.
Why It Matters
Your access to free AI tools likely stays the same, but new models might take longer to reach you.