A developer in Hangzhou ships a model on Monday and wonders which of a dozen overlapping rules actually applies. A hospital in Shanghai hesitates to trial a diagnostic system because no single law tells it who is liable when the model is wrong. A legislator in Beijing reads the same complaints and decides the patchwork has gone on long enough.
China does not yet have one law called the "AI law (人工智能法)." What it has is a thicket of separate rules written for separate moments: one for recommendation algorithms, one for deepfakes, one for generative chatbots. The 2026 legislative calendar changes that. Both the State Council's annual legislative work plan and the National People's Congress (NPC) Standing Committee's plan now name a comprehensive artificial-intelligence law as a live item — not yet a bill on the floor, but officially on the desk.
What actually got scheduled
In the State Council 2026 Legislative Work Plan (国务院2026年度立法工作计划), issued in May 2026, the relevant line reads: improve AI governance and "accelerate comprehensive legislation for the healthy development of artificial intelligence (加快推进人工智能健康发展综合性立法)." The NPC Standing Committee's 2026 plan goes further in process terms, listing AI-related legislative projects as a preparatory review item (预备审议项目) and instructing relevant bodies to "press ahead with research and drafting, and arrange deliberation as circumstances allow."
That wording matters. A preparatory review item is not a law about to pass; it is the stage where a topic is studied, drafted, and held ready. It signals political commitment and a working timeline without fixing an enactment date. No effective date for a comprehensive AI law exists yet, and any claim that one has been set would be false.
Why now, and why a single law
The existing framework is real but scattered. China's AI governance is often described by scholars as a "1+3+2+N" structure:
- 1 — the Civil Code (民法典), which sets baseline civil rights and obligations.
- 3 — three foundational statutes: the Cybersecurity Law (网络安全法), the Data Security Law (数据安全法), and the Personal Information Protection Law (个人信息保护法).
- 2 — two local comprehensive regulations, in Shanghai and Shenzhen.
- N — a growing stack of sectoral rules, including the Interim Measures for the Management of Generative AI Services (生成式人工智能服务管理暂行办法) from August 2023, the Measures for the Labeling of AI-Generated Synthetic Content (人工智能生成合成内容标识办法) from September 2025, and newer rules on anthropomorphic interaction and autonomous agents (智能体).
Each rule solved a pressing problem when it appeared. Together they created overlap, gaps, and uncertainty about which agency leads. A comprehensive law is meant to sit above the sectoral rules and define who regulates what — the so-called "multi-head regulation (多头监管)" problem that lawmakers themselves have named.
The groundwork was laid before the law
The move is not a leap. In October 2025, the NPC Standing Committee amended the Cybersecurity Law (网络安全法) to add dedicated clauses on AI, marking the first time a foundational statute named AI governance directly. Those clauses took effect on January 1, 2026. Academic drafts have been circulating for years: the Chinese Academy of Social Sciences published a model AI law (示范法) in 2023, updated to version 3.0 by 2025, and seven research institutions released a scholars' draft in 2024.
At the March 2026 press corridor (部长通道) of the NPC session, the Minister of Justice responded directly to questions about AI legislation, framing it as both a development and a risk-management priority. The public signal is consistent: encourage innovation, but write the boundaries down.
What the law is expected to cover
Based on published drafts and official statements, the comprehensive law is likely to address:
- Accountability — who is responsible when an AI system causes harm, especially in medical and educational use.
- Training-data obligations — sourcing, quality, and intellectual-property compliance, building on the standards already in force.
- Sectoral coordination — how the existing generative-AI, deep-synthesis (深度合成), and labeling measures fit under one statute.
- Innovation posture — several legal scholars have argued for a "weak regulation, strong promotion (弱规制、强促进)" approach that keeps barriers low while fixing safety floors.
None of this is settled text. The drafts disagree on scope, and the final law will almost certainly differ from any current version.
Why this matters beyond China
A consolidated Chinese AI statute is being watched as a possible template. The EU's AI Act and China's emerging law take different routes — the EU legislates horizontally by risk tier, while China is folding a mature stack of sectoral rules into a single frame. For foreign firms operating in China, the practical effect of consolidation is predictability: fewer conflicting rules, one accountability map.
What readers can do now
- Track the NPC Standing Committee's annual legislative plan each spring; a preparatory review item that moves to the formal agenda is the earliest reliable signal of momentum.
- If you build or deploy AI in China, audit your compliance against the existing sectoral rules now — the comprehensive law is expected to absorb, not replace, them.
- Read the published scholars' drafts (the CASS model law, version 3.0) as the closest available preview of the legislature's likely vocabulary and structure.
Honest limitations
This article is based on the published State Council and NPC Standing Committee 2026 legislative work plans, the October 2025 Cybersecurity Law amendment, and academic drafts reported in Chinese legal media. Status is "in drafting / preparatory review" — no comprehensive AI law has been enacted, and no effective date has been announced. Descriptions of the law's likely contents are inferences from drafts and official commentary, not from enacted text. Readers needing legal certainty should consult the official plan documents and qualified counsel.