China wants mandatory labels on all AI-generated content posted online
China's internet regulatory body, the Cyberspace Administration of China (CAC), has proposed a new set of guidelines that would mandate digital platforms to clearly mark content produced by artificial intelligence (AI). The draft plan was announced on Saturday and is currently open for public comments until October. The proposed rules require all AI-created material to be identified with a visible logo and embedded metadata within the relevant files.
Proposed guidelines for different media formats
The draft plan suggests that the identifying logos should be placed in multiple locations within a text, image, audio, or video file. For audio files specifically, a voice prompt is proposed to notify listeners of AI-generated content at the start and end of a file. Video players would only need to display notices about AI content at the beginning and end of a clip.
User responsibilities and platform requirements
Under the proposed regulations, users who post AI-generated content will be obligated to label it as such. If they utilize generation tools provided by a platform, they'll need to identify themselves and their activities will be logged for six months. In cases where metadata is missing from a file, service providers will have to analyze content and label it if they suspect it was created by AI.
Potential penalties and information sharing
The draft code also outlines potential penalties for platform operators and individuals who fail to comply with these requirements. Additionally, it encourages China's internet platforms to share their insights as they develop methods to detect AI content, aiming to improve collective efforts in this area. This proposal is consistent with China's broader approach of requiring tech giants to monitor online content and link user-generated material to an identifiable individual.