The Cyberspace Administration of China hasAI.

Recently, the Central Cyberspace Administration of China launched the second phase of the special action "Clear and Bright: Rectifying the Chaos in AI Applications." This initiative focuses on prominent issues such as the use of AI to create and disseminate false information, spread violent and vulgar content, impersonate others, and infringe upon the rights of minors. Local cybersecurity departments have been tasked with advancing key rectification tasks and supervising websites and platforms to concentrate on clearing various types of illegal and harmful AI-generated information. Violating accounts and mobile applications are being investigated according to the law, effectively curbing the chaos of AI-generated synthetic information and promoting the healthy and orderly development of AI applications. So far, over 5.61 million illegal and inappropriate pieces of information have been cleared, more than 49,000 accounts investigated, and over 2,400 websites and applications addressed. Since the commencement of the special action, local cybersecurity departments have actively fulfilled their responsibilities in regulatory oversight, implementing a series of targeted governance measures suited to local conditions. The Beijing Cybersecurity Administration has guided platforms to comprehensively upgrade their review strategies, continuously enhancing their capacity to identify and address new variants of violations. The Shanghai Cybersecurity Administration has urged platforms to carry out in-depth self-examinations and increase the intensity of investigations into illegal information and accounts, launching an AI application "Knowledge Awareness Package" to help the public recognize potential risks. The Zhejiang Cybersecurity Administration has provided compliance guidance through a "One Policy for One Enterprise" approach for key local platforms, urging improvements to internal control systems and handling a number of problematic AI application software in accordance with the law. The Guangdong Cybersecurity Administration has held work meetings with key websites and platforms to directly supervise the rectification of various application violations, ensuring the special action achieves tangible results. Websites and platforms have actively responded to the deployment arrangements of the special action, continuously improving the normalization governance mechanism for AI content. Platforms such as Douyin, Kuaishou, Weibo, Tencent, Baidu, Bilibili, Xiaohongshu, Zhihu, Douban, and Taobao have strengthened technical support, optimized multimodal recognition models, dynamically expanded face, voiceprint, and violation sample libraries, and launched real-time face governance systems, among other functional measures, to enhance the efficiency of intelligent detection and interception of inappropriate content. They have intensified efforts to handle violations related to accounts and information, ensuring smooth channels for reporting leads, and have published a total of 46 governance announcements. Platforms like Doubao, Yuanbao, Qianwen, and Wenxin Yiyan have strictly overseen the review of original corpus content, limited the output of inappropriate content, and implemented requirements for labeling AI-generated synthetic content to prevent illegal and inappropriate information from being generated and disseminated from the source. Application stores such as Huawei, Xiaomi, OPPO, and Vivo have established strict entry standards and listing requirements for APP developers, conducting random inspections and reviews of listed APPs to effectively intercept illegal applications from entering the system. During the special action, a number of illegal accounts and applications have been dealt with. To further strengthen warning education, some typical cases are reported below: 1. Using AI to "magic modify" classics, generating "digital refuse." Weibo account "Tian**Xiao," Douyin account "Wai**Guo," Kuaishou account "Xiao*Jun," WeChat account "Xian*i," etc., abused AI technology to "magic modify" classics such as "Romance of the Three Kingdoms" and "Journey to the West," subverting the character setups of important figures and fabricating low-quality bizarre plots to gain traffic. The relevant accounts have been processed in accordance with the law. 2. AI generating and publishing false information. Kuaishou account "Yun**Shi," Xiaohongshu account "You*Tou," Douyin account "7**Yuan," WeChat account "Chang**Fan," etc., utilized AI to create synthetic false disaster scenes and rescue footage, concocted false social hot topics, published feudal superstitious content, and used AIDIGIHUMAN for false advertising, misleading the elderly and inciting panic. The relevant accounts have been processed in accordance with the law. 3. AI impersonation of others. Douyin account "Te**Li," Kuaishou account "Shi*," Xiaohongshu account "Chao**Jian," WeChat account "M**Le," etc., employed AI techniques such as "face swapping" and "voice swapping" to impersonate public figures, generating synthetic images and videos of vulgar dances, hugs, and kisses for marketing profit. The relevant accounts have been processed in accordance with the law. 4. Creating and publishing violent and vulgar content. Bilibili account "Dai**Zhe," Douyin account "Bo**C," Kuaishou account "Xia**Gui," WeChat account "Xing**7," etc., have repeatedly published AI-generated synthetic content containing terrifying, bizarre, violent, and vulgar images. The relevant accounts have been processed in accordance with the law. 5. AI infringing upon the rights of minors. Xiaohongshu account "Xiao**Wa," WeChat account "Dou*a," Douyin account "Ban**Li," Kuaishou account "Ruo**Xue," etc., published AI-generated synthetic content depicting minors inappropriately pregnant or fighting, batch-producing and disseminating "toxic animations" and children's "abuse-oriented" or "cult" videos that harm and mock minors images, affecting their physical and mental health. The relevant accounts have been processed in accordance with the law. 6. Using AI "agency" to engage in internet water army activities. WeChat account "Tao**Ke," Weibo account "A**Shi," Douyin account "Feng*," etc., directed traffic for AI "agency" software, simulating real-person-operated accounts and interaction for generating a large volume of low-quality, homogeneous content. The relevant accounts have been processed in accordance with the law. 7. AI product services and application violations. Multiple intelligent agents on websites such as "N**t," "Chuan**Yuan," "Tian*Quan," "Yu**o," contain obscene, pornographic, or other harmful content to the physical and mental health of minors in covers, introductions, and generated dialogues, severely disrupting the online ecosystem. Those refusing to correct violations will be dealt with according to the law, with measures including warnings, fines, delisting, or closure taken as appropriate. Going forward, cybersecurity departments will summarize the achievements of the special action, further grasp the trends and laws of AI development, keep a close watch on new situations and risks related to AI-generated synthetic content, and timely take effective measures to prevent the abuse and misuse of AI technologies, continuously creating a favorable environment for the innovative development of artificial intelligence. This article is reproduced from the Cyberspace China public account, GMTEight editor: Chen Siyu.