Mei Ri Jing Ji Xin Wen·2025-09-03 11:08Core Points - The implementation of the "Artificial Intelligence Generated Synthetic Content Identification Measures" aims to ensure that all AI-generated content must be clearly identified to maintain the authenticity and transparency of information dissemination [2] - The practical execution of the new regulations faces significant challenges, particularly in the areas of explicit identification and automatic recognition by major content platforms [2][15] Generation Side Challenges - Explicit identification of AI-generated text content is largely absent, with all tested AI applications failing to add any form of explicit identification as required by the regulations [3] - In contrast, image content generated by the tested applications generally complied with the regulations, adding clear text prompts or brand watermarks [9] - Some video generation tools offer "watermark removal" as a paid membership service, allowing users to eliminate the required AI content identification [10][14] Distribution Side Issues - Major content platforms such as Xiaohongshu, Douyin, and Weibo failed to automatically recognize and add identification to AI-generated content during testing [15] - Douyin acknowledged limitations in its technical capabilities to verify and identify AI-generated works, while Weibo and Xiaohongshu did not provide direct responses regarding their failures to recognize AI content [16][15] Technical Challenges and Future Directions - The fragmentation of identification embedding and detection technologies poses a significant challenge, making it difficult to establish a unified compliance detection system [19][21] - Future efforts will focus on developing a standardized identification technology and enhancing implicit identification methods, such as digital watermarks that are not easily removed [22] Industry Response and Collaboration - Following the implementation of the regulations, various applications and platforms have begun to announce their compliance measures and technical solutions [24] - Companies like MiniMax and Xiaohongshu are collaborating to form a unified standard for implicit identification, emphasizing that content identification can promote innovation rather than hinder it [24][25]