China brings into force Regulations on the Administration of Deep Synthesis of Internet Technology addressing deepfakes and similar technologies

The Regulations aim to address the risks related to so-called “deep synthesis”, an AI-based technology that enables content synthesis and the creation of virtual digital “humans” online. Content created by deep synthesis is often highly realistic, such as in cases with changing or augmenting facial features, and increasingly confusing to the general public. The highly realistic outputs, ease of operation and low cost create potential safety and security risks for the deep synthesis technology, as it can be used by criminals to produce, copy and disseminate illegal or false information or assume other people’s identities to commit fraud. 

The Regulations impose obligations on the providers and users of so-called “deep synthesis technology” (deep learning, machine learning and other algorithmic processing systems) which uses mixed datasets and algorithms to produce synthetic content, such as deepfakes. The Q&A for the Regulations list a wide number of technologies as being covered, such as technologies that generate or edit text content (e.g. text generation and chat bots), video and audio (e.g. speech generation, face and posture manipulation in video services and image enhancement and generation), virtual scene generation and 3D reconstruction.

Under the Regulations, deep synthesis providers must, among other things: 

• prevent the use of deepfake services to produce, copy, publish or disseminate information prohibited by laws and administrative regulations, or engage in other activities prohibited by laws and administrative regulations;
• ensure that their services fully comply with applicable laws and regulations, including the Personal Information Protection Law, the Data Security Law and the Cybersecurity Law;
• establish and maintain systems for user registration and verification of user identity, provide reviews and ethical evaluations of the deepfake services and the algorithms used by the system;
• where a deep synthesis system provides significant editing functions for biometric information, such as faces and voices, the provider should prompt the user to notify the individual whose personal information is edited and obtain their separate consent; 
• implement measures to address data security, personal information protection, telecom fraud, emergency response and other technical safeguards, and record and retain relevant network logs;
• develop, implement and publish the rules and terms and conditions for the use of their platform; 
• adopt content moderation rules, as well as automated and human methods for reviewing the input data and synthesized results to identify illegal or harmful information; and
• implement notice and takedown procedures to deal with the publication of false, illegal or harmful information by deep synthesis users, promptly take measures to dispel rumours, restrict and terminate accounts, suspend deep synthesis services and report inappropriate use to relevant authorities.

Deep synthesis providers must also ensure that labels are added to the content generated using deep synthesis technology and conduct reviews of their algorithms and security procedures.

Read he Regulations here and the FAQs here (both only in Chinese).