China

The Measures apply broadly to the development and use of GenAI technology to provide services for the generation of text, pictures, audio, and video content to the public within Chinese territory. The Measures overlap substantially with the earlier-promulgated regulations in terms of content moderation, cybersecurity compliance and algorithm regulation.

The regulations apply to “deep synthesis” (a subset of gen AI technologies, i.e., technology using deep learning, virtual reality, and other synthetic algorithms to create text, images, audio, video, virtual scenes or other information) to provide internet information services in the PRC.

These provisions regulate the use of algorithmic recommendation technology to provide Internet information services within the mainland China, including applying algorithmic recommendation technologies to create content, make personalized recommendations, rank or select information, search or filter content, dispatch service providers, or otherwise provide information to users.

These provisions apply to online services “capable of creating public opinions or social mobilization” including for example forums, blogs, chat rooms, public accounts, etc. All AI-specific regulations require security assessment in accordance with these Security Assessment Provisions for Public Opinion Services if the corresponding AI services are “capable of creating public opinions or social mobilization”.

Shanghai and Shenzhen were pioneer cities in China adopting their own policies to facilitate AI industries, followed by other local governments in Beijing, Zhejiang, Jiangsu, Guangdong, etc. Unlike existing AI-related regulations at the national level, these local government regulations are essentially strategic papers and action plans aiming to promote the development of the AI industry and strengthen the functions of new generation AI science and technology (S&T) innovation sources. These local policies mainly set out high-level principles, from a government perspective, that demonstrate the position to “encourage”, “support” and “promote” the AI industry.

A national strategy released by the State Council of China that is aimed at embedding AI across China’s economy. It mandates integration in six arenas—science and technology, industry, consumption, people’s well-being, governance, and global cooperation—to reshape production and daily life and accelerate an “intelligent economy” and “intelligent society.”

This Framework is a government playbook that explains how to rate AI risks, label and trace AI-generated content, run safety checks before and after launch, set duties for open-source and service providers, and apply extra rules in sensitive sectors. Its goal is AI that stays safe, under human control, and accountable across industries, while coordinating internationally on common standards.

The Chinese government’s AI Action Plan lays out a framework for international cooperation to ensure safe, inclusive, and development-oriented AI under principles such as sovereignty, fairness, and open collaboration.

These basic requirements, issued by China’s National Cyber Security Standardization Technical Committee, aims to clarify the basic requirements for generative AI services in terms of security, including in particular corpus security, model security, security measures, etc., and also sets out security assessment requirements.

The TC260 Practice Guide contains guidelines for content identification in genAI services to comply with regulations and policies, including the Interim Measures for the Administration of Generative Artificial Intelligence Services. The practice guide provides methods for marking generated content in the four categories of text, image, audio, and video, including explicit and implicit watermark identification.

Measures jointly issued by the Cyberspace Administration of China (CAC), the Ministry of Industry and Information Technology, the Ministry of Public Security, and the State Administration of Radio and Television, and intended to further standardize the identification activities of synthetic content generated by AI. FAQs issued by CAC.

These principles and codes are not statutory requirements, but mainly ethical guidance aiming to emphasize that development of AI technologies should generally adhere to the principles of “harmonious and friendly, fair and just, inclusive and open, respectful of privacy, secure and controllable, shared responsibility, open and collaborative, and agile governance”.