Herbert Smith Freehills
View original resourceHerbert Smith Freehills' comprehensive tracker provides expert legal analysis of China's groundbreaking AI law proposal introduced to the National People's Congress in June 2025. This isn't just another regulatory summary—it's a living resource that dissects what could become the world's second major AI governance framework after the EU AI Act. The tracker breaks down the proposed risk-based classification system, maps out compliance obligations across the AI value chain, and analyzes how this legislation could reshape global AI governance. For organizations with operations in China or those supplying AI technologies to Chinese markets, this resource is essential reading as the proposal moves through the legislative process.
Unlike the EU AI Act's prohibition-based approach, China's proposed framework emphasizes innovation enablement alongside regulatory oversight. The law introduces a unique three-tier risk classification system that distinguishes between "basic," "significant," and "high-risk" AI applications—with notably different criteria than EU classifications. The proposal also establishes mandatory ethics impact assessments for certain AI deployments and creates specialized governance bodies within existing regulatory structures. Perhaps most significantly, the law includes provisions for cross-border data governance that could affect international AI companies differently than domestic Chinese developers.
Current status: Under review by NPC Standing Committee following June 2025 introduction
Expected milestones:
The tracker provides regular updates as the proposal moves through China's legislative process, including amendments, committee feedback, and industry consultation responses. Herbert Smith Freehills also analyzes how geopolitical factors and international AI governance developments might influence the final legislation.
Financial services: Algorithmic trading and credit scoring systems face enhanced transparency requirements and mandatory algorithmic audits every 18 months.
Healthcare: AI diagnostic tools require pre-deployment safety assessments and ongoing performance monitoring, with specific provisions for medical AI data retention.
Autonomous vehicles: Three-phase testing framework with escalating requirements for urban deployment, including mandatory insurance provisions.
Social media and content platforms: Content recommendation algorithms must implement "explainability features" and allow users to access alternative algorithmic experiences.
Manufacturing and logistics: Industrial AI systems face lighter regulatory burdens but must comply with worker notification requirements when AI systems make employment-related decisions.
Once enacted, organizations should prepare for these anticipated timelines:
The tracker includes downloadable compliance checklists and regularly updated regulatory calendar to help organizations plan their implementation strategies.
Published
2025
Jurisdiction
China
Category
Regulations and laws
Access
Public access
VerifyWise helps you implement AI governance frameworks, track compliance, and manage risk across your AI systems.