Correction (19 October 2026): We corrected our description of the Chinese rules: the security assessment and algorithm filing apply only to generative AI services with public-opinion attributes or social-mobilisation capacity, the content-marking and identity-verification duties are now stated precisely, and data-localisation rules apply only in defined cases.

Led by the Cyberspace Administration of China (CAC), this body of regulation focuses less on the technology itself than on content control, social stability, and state security.

📊

Algorithmic Recommendation Systems Provisions (2021)

The world's first algorithm law. It prohibits algorithms that manipulate users, create addiction, or engage in price discrimination on social media and e-commerce platforms. It requires companies to file their algorithms with the state through the Algorithm Registry.

🎭

Deep Synthesis Provisions (2022)

Targets deepfake and synthetic content creators. It requires providers to add technical marks to AI-generated or edited content, to apply prominent labels where synthetic content may confuse or mislead the public, and to verify the real identity of the service's users.

🧠

Generative AI Interim Measures (2023)

The world's first binding regulation for LLMs (Large Language Models) like ChatGPT. It requires outputs to align with "Core Socialist Values," and services with public-opinion attributes or social-mobilisation capacity must undergo a state security assessment and file their algorithms (Art. 17).

The "Data and Content" Barrier to Entering the Asian Market

To launch a generative AI service in the Chinese market, a provider must use lawfully sourced training data that does not infringe intellectual-property rights (Art. 7) and keep its outputs within the state's content rules. Data-localisation and export-assessment rules also apply where critical information infrastructure, important data or large volumes of personal information are involved.

Why Should Global Companies Pay Attention?

China's regulatory model does more than protect its domestic market; developing countries across Southeast Asia, the Middle East, and Africa also use it as a template when drafting their own AI laws. Building the Chinese compliance mechanisms (especially the algorithm transparency and watermarking requirements) into a system helps a company prepare in advance for the regulatory pressure it will face elsewhere.

Review the Official Legislation Text

The official text of the "Interim Measures for the Management of Generative Artificial Intelligence Services" (2023) , published by the Cyberspace Administration of China (CAC), is available at the link below.

Generative AI Measures Legislation (CAC.gov.cn)
📄

Original documents

Read the primary sources referenced in this article:

↓ Interim Measures for Generative AI Services — text & official CAC source