On 25 November 2022, three Chinese authorities jointly issued the Provisions on the Administration of Deep Synthesis of Internet-Based Information Services: the Cyberspace Administration of China, the Ministry of Industry and Information Technology, and the Ministry of Public Security. The rules took effect on 10 January 2023 and were among the first national regulations anywhere aimed specifically at synthetic media, or “deepfakes.”
The provisions define deep synthesis as technology that uses machine learning and other techniques to generate or edit text, images, audio, video, and virtual scenes. Service providers must add labels or tags to content produced with deep synthesis, with particular attention to content involving voice simulation, AI-generated writing imitating real people, face synthesis, and face manipulation, so the public is not misled. Providers must authenticate the real identity of their users, establish mechanisms to identify and handle false or harmful synthetic material, keep records, and report violations to the authorities. Platforms with significant public influence must also complete algorithm filing procedures and display their filing numbers.
These rules placed China ahead of most jurisdictions in mandating disclosure of synthetic content, anticipating concerns that the EU AI Act and various US measures would address later. For businesses operating in China, they established that generating or distributing AI-made media carries explicit labeling and identity-verification duties, with the labeling concept now spreading to other legal systems.