China's AI Governance: Generative AI Measures & Algorithm Registry
How China regulates AI: algorithm registration, generative AI service rules, deep synthesis requirements, and content filtering mandates.
Last updated: March 2026
Overview
China has taken a rapid, layered approach to AI regulation, issuing multiple targeted regulations rather than a single comprehensive framework. These regulations focus heavily on content control, algorithmic transparency, and state security.
Key Regulations
Generative AI Measures (2023)
Interim measures governing generative AI services. Requires providers to ensure content aligns with "socialist core values," implement content moderation, and conduct security assessments before launch.
Algorithm Recommendation Rules (2022)
Requires companies to register algorithms with the Cyberspace Administration of China (CAC). Users must be able to opt out of personalized recommendations.
Deep Synthesis Provisions (2023)
Regulates deepfakes and AI-generated content. Requires watermarking, user identity verification, and content labeling for all "deep synthesis" outputs.
Key Requirements for AI Providers
- Register algorithms with the CAC before public deployment
- Conduct security assessments for generative AI services
- Implement real-time content filtering aligned with state regulations
- Label all AI-generated content with visible watermarks
- Maintain training data records for government inspection
- Ensure training data does not contain content that undermines state power