China publishes the Interim Measures for the Management of Generative AI Services

Background

In 2023 China's regulators set out rules for generative AI services offered to the public. The measures were issued jointly by seven government bodies, led by the Cyberspace Administration of China (CAC).^[1][2]

What happened

On 13 July 2023 the CAC published the Interim Measures for the Management of Generative Artificial Intelligence Services (生成式人工智能服务管理暂行办法). They were issued as CAC Order No. 15 and dated 10 July 2023.^[1] The seven issuing bodies were the CAC, the National Development and Reform Commission, the Ministry of Education, the Ministry of Science and Technology, the Ministry of Industry and Information Technology, the Ministry of Public Security and the National Radio and Television Administration.^[2]

The CAC adopted the measures at its 12th office meeting on 23 May 2023. They took effect on 15 August 2023.^[1]

Key details

The text has 24 articles in five chapters: general provisions; technology development and governance; service standards; supervision, inspection and legal liability; and supplementary provisions.^[1]

Article 2 sets the scope. The measures cover services that use generative AI to provide text, images, audio, video and other content to the public within mainland China. Research and development by companies, universities and research institutions that do not offer services to the domestic public is excluded.^[1] Reuters reported that the final rules applied only to services offered to the public, as regulators sought to support development of the technology.^[3]

The measures followed draft measures published on 11 April 2023. The final version was substantially amended from the draft to be more friendly to commerce, research and innovation.^[4]

Article 7 covers training data. Providers must use training data and base models from lawful sources, respect intellectual property, obtain consent or another legal basis for personal information, and take measures to improve the authenticity, accuracy, objectivity and diversity of training data.^[1] A base model is a large model trained in advance on broad data, and other applications are often built on top of it.

Article 12 requires providers to label generated images and videos under China's deep synthesis rules.^[1] Article 17 requires services with public-opinion attributes or social mobilisation capacity to undergo a security assessment and complete algorithm filing.^[1]

Article 20 concerns generative AI services provided into China from abroad. If such a service violates the measures, the national cyberspace authority is to notify the relevant bodies so that they can take technical and other necessary measures.^[1]

Significance

On 31 August 2023 the first chatbots cleared under the new regime opened to the general public, led by Baidu's ERNIE Bot (文心一言).^[5] AFP reported that SenseTime (商汤科技), Baichuan Intelligent Technology (百川智能) and Zhipu AI (智谱 AI) also received approval for their generative AI apps in the same round.^[6]

By 31 December 2025, 748 generative AI services had completed national filing and 435 generative AI apps or features had completed local registration. A further 446 filings were added during 2025.^[7] By 31 August 2026 the cumulative total had reached 1112 filed services and 731 registered apps or features.^[8]

For readers, the measures set the basic rules that services offered to the Chinese public must follow, and the filing counts show how many services have gone through that process since the rules took effect.