Algorithmic Regulation in China

China has developed one of the world’s most comprehensive regulatory frameworks for algorithms and artificial intelligence. The regulatory landscape is defined by three principal instruments: the Internet Information Service Algorithmic Recommendation Regulation (2022), the Deep Synthesis Regulation (2023), and the Interim Measures for the Management of Generative AI Services (2023). These regulations are administered by the Cyberspace Administration of China (wangxinban, 网信办).

Algorithmic Recommendation Regulation (2022)

The Administrative Provisions on Algorithmic Recommendations in Internet Information Services, effective 1 March 2022, were the first comprehensive regulation of algorithmic recommendation systems. The Provisions apply to all internet information service providers that use algorithms to recommend content — including news feeds, search results, short videos, and product recommendations.

The Provisions establish several core obligations. Algorithmic recommenders must publish the basic principles, purpose, and main operating mechanisms of their algorithms. They must provide users with options to disable algorithmic recommendations or select non-personalized alternatives. Algorithmic recommenders that provide news information services must obtain relevant licenses and ensure that recommended content complies with state-approved sources. The Provisions prohibit algorithmic recommendation that promotes harmful content, including content that subverts state power, undermines national unity, or endangers public order.

The Provisions also regulate algorithmic price discrimination — prohibiting the use of algorithms to set differential prices based on user transaction characteristics, browsing history, or other behavioral data. They require protection of user data and prohibit algorithms that facilitate fraudulent activities or infringing content. Service providers must conduct security assessments of their algorithms and file assessments with the CAC.

Deep Synthesis Regulation (2023)

The Administrative Provisions on Deep Synthesis in Internet Information Services, effective 10 January 2023, regulate the use of deep learning and virtual reality technologies to generate or alter content (shen du he cheng, 深度合成). The provisions apply to any technology that creates, substitutes, or manipulates images, video, audio, or text to produce synthetic content that may be indistinguishable from authentic content.

Providers of deep synthesis services must label generated content, ensuring that users can identify synthetic content as artificial. They must establish management systems for training data and prevent the generation of illegal content. Deep synthesis providers are prohibited from generating content that violates socialist core values, endangers national security, or infringes the rights of others. They must implement real-time content filtering and maintain records of generated content for at least three months.

The Regulation specifically addresses voice and facial synthesis, requiring explicit consent from individuals whose voice or facial data is used. It prohibits the generation of fake news and the use of deep synthesis for fraud or defamation. Service providers must conduct security assessments before launching deep synthesis services and must report security incidents to the CAC.

Generative AI Measures (2023)

The Interim Measures for the Management of Generative AI Services, effective 15 August 2023, represent the world’s first comprehensive regulation of generative artificial intelligence. The Measures apply to providers of generative AI services to the public within China, including large language models, text-to-image systems, and text-to-video generators.

Core requirements include: content compliance with socialist core values; truthful and non-infringing training data; non-discrimination obligations; labeling of AI-generated content; and security assessment filing. Providers must prevent the generation of content that subverts state power, undermines national unity, advocates terrorism or extremism, promotes ethnic hatred, or spreads obscenity. Training data must be sourced from legitimate channels and must not infringe intellectual property rights.

The Measures require providers to implement mechanisms for user registration, content filtering, and complaint handling. Providers are legally responsible for content generated through their services and must cooperate with government investigations. Violations may result in warnings, fines, service suspension, or license revocation.

Algorithmic Registration and Transparency

A key innovation of China’s algorithmic regulation is the algorithmic registration system. Major algorithmic recommendation services must register their algorithms with the CAC and file detailed descriptions of their algorithmic mechanisms, data processing practices, and risk management measures. The registration information is published on the CAC’s website, providing public transparency about the algorithms that shape China’s information environment.

The registration requirement applies to algorithms used for news recommendation, search ranking, content distribution, and other high-impact applications. Registrants must disclose the basic principles and purpose of the algorithm, the types of data used, the scope of application, and the measures taken to prevent algorithmic harm. The system creates a regulatory record that the CAC can use for supervision and enforcement.

Wangxinban Oversight

The Cyberspace Administration of China (wangxinban) is the primary regulator of algorithms in China. Established in 2014 and elevated to ministerial status in 2022, the CAC exercises comprehensive authority over internet information services, including algorithmic regulation. The CAC issues administrative regulations, conducts security assessments, investigates violations, and imposes administrative penalties.

The CAC has established specialized departments for algorithmic regulation and AI governance. It coordinates with other regulators, including the Ministry of Industry and Information Technology and the State Administration for Market Regulation, to ensure consistent enforcement. The CAC also participates in international AI governance discussions, promoting China’s approach to AI regulation as an alternative to Western models.