For a country that moves fast on technology regulation, China's plan for a single, comprehensive AI law has been conspicuously slow. The bill has sat on the legislative agenda for years, yet no unified statute exists — and as of late 2026 the draft has still not been sent to the nation's top legislature for formal review.
This is not a fringe proposal. "人工智能法" (the Artificial Intelligence Law) has been name-checked in State Council and National People's Congress (NPC) planning documents since 2023. But planning and enacting are different things, and the gap between them is the story.
Key takeaways
- China has no unified AI basic law — the comprehensive statute remains in the drafting and study stage.
- It is not in effect and not even formally under review; the NPC lists it as a "preparatory review" item.
- Regulators instead govern AI through sectoral rules — a deliberate "small-cut" strategy.
- Officials keep signaling urgency, but the wording has shifted from a concrete "draft" toward a vaguer "accelerate the legislative process."
How the law climbed the agenda, then stalled
The trajectory is visible in the annual legislative work plans. In 2023, a document described as the "Artificial Intelligence Law draft" (人工智能法草案) first appeared as a project预备提请全国人大常委会审议 — slated for preliminary submission to the NPC Standing Committee. In 2024 the same framing returned.
Then the language softened. The 2025 State Council plan dropped the word "draft" and spoke instead of "advancing legislative work for the healthy development of AI" (推进人工智能健康发展立法工作). By 2026 the tone sharpened again: the State Council's 2026 legislative work plan (国办发〔2026〕14号, issued May 8, 2026) called for "improving AI governance and accelerating comprehensive legislation for the healthy development of AI." The NPC Standing Committee's 2026 plan likewise placed AI healthy-development legislation on its preparatory-review list.
On July 31, 2026, a National Development and Reform Commission (国家发展改革委) spokesperson publicly said the agency would "accelerate the legislative process of the AI law" (加快人工智能法的立法进程). Yet, as multiple legal analyses note, no official draft had been submitted to the NPC Standing Committee for review as of August 2026. The law remains in the drafting-and-deliberation phase.
Why Beijing prefers the patchwork
In the absence of a master statute, China governs AI through a stack of narrower instruments, each aimed at a specific risk:
- The Interim Measures for Generative AI Services (生成式人工智能服务管理暂行办法), in effect since August 15, 2023.
- The Provisions on Deep Synthesis (深度合成管理规定), covering face-swapping and voice-cloning.
- The Algorithm Recommendation Management Provisions (算法推荐管理规定), governing feeds and recommendations.
- The Measures for Labeling AI-Produced and Synthetic Content (人工智能生成合成内容标识办法), in force since September 1, 2025.
The logic is to regulate the most urgent, well-understood risks first, through administrative rules that can be revised quickly, rather than locking everything into one slow, rigid law. According to Li An, chief scientist at BrainNet (脑机网), this "small-cut" approach lets regulators learn from practice before committing to a binding framework.
What "preparatory review" actually means
The NPC's "preparatory review" category (预备审议项目) is precisely for bills whose conditions are not yet ripe — projects that need more study and argument before they can enter formal deliberation. Being on that list is a signal of intent, not a step toward enactment. So any claim that China has "enacted" or "passed" an AI law is wrong; the correct status is draft / not yet enacted.
How this compares internationally
The contrast is sharp. The European Union's AI Act has been moving through staged implementation, giving Brussels a cross-sectoral statute years ahead of Beijing's comprehensive bill. China, by contrast, chose to ship targeted rules first and defer the umbrella law. That sequence reflects a calculated bet: govern what you can see today, and write the grand framework once the technology and its risks are clearer.
What readers can do now
- If you comply with Chinese AI rules, do not wait for the AI law. The binding obligations today are the sectoral measures already in force — generative-AI service rules, deep-synthesis provisions, algorithm filing, and content labeling.
- Track the legislative signals, not the rumors. Watch the annual State Council and NPC plans; a shift from "advance the work" back to a named "draft" would be the real tell that enactment is near.
- Build compliance that survives a future law. Because the patchwork already covers labeling, algorithm filing, and training-data duties, a robust program today will likely satisfy tomorrow's statute.
Honest limitations
This article describes the status of China's comprehensive AI law as of late 2026, based on State Council and NPC legislative work plans and secondary legal analyses. I verified that no unified AI law has been enacted and that the bill remains in the drafting/preparatory-review stage; the July 31, 2026 NDRC statement and the May 8, 2026 State Council plan (国办发〔2026〕14号) are the anchor facts. I did not obtain the text of any internal draft, and I cannot predict the enactment date. The comparison with the EU AI Act is framed at a high level and does not detail every provision. The Li An line is a general expert observation, not a verifiable statistic.