The Cyberspace Administration of China recently announced the latest achievements of the second phase of the "Clear and Bright - Rectifying AI Application Chaos" special operation. It has carried out in-depth governance against prominent issues such as using artificial intelligence technology to create and spread false information, disseminate violent and vulgar content, impersonate others, and infringe on the rights and interests of minors. As of now, a total of more than 5.61 million illegal and irregular pieces of information have been cleaned up nationwide, 49,000 illegal accounts have been investigated, and over 2,400 illegal websites and applications have been dealt with, effectively curbing the initial signs of AI technology being misused and maliciously controlled.

In this special operation, network administration departments in various regions have introduced precise and efficient governance measures based on their actual conditions. Beijing has guided major platforms to upgrade their content review strategies, significantly enhancing their ability to identify and handle new variants of illegal content; Shanghai has urged companies to self-inspect and correct issues, and launched an "AI application knowledge package" to guide the public to enhance their awareness of prevention; Zhejiang has implemented "one enterprise, one strategy" compliance guidance for key platforms within its jurisdiction, and lawfully investigated a number of AI applications with prominent problems; Guangdong has held regular meetings with key platforms to conduct targeted urging for the rectification of violations in multiple applications. At the same time, mainstream platforms such as Douyin, Kuaishou, Weibo, Tencent, Baidu, Bilibili, Xiaohongshu, Zhihu, Douban, and Taobao have continuously strengthened technological empowerment, improving multi-modal recognition models, dynamically expanding face and voiceprint libraries, and launching real-time face governance systems, which greatly improved the efficiency of intelligent detection of illegal content, during which they released 46 related governance announcements.

Regarding source prevention, leading large model platforms such as Douyin, Yuanbao, Qianwen, and Wenyiyan have strictly controlled the review mechanism of original text content, and fully implemented the labeling requirements for AI-generated content, preventing the generation and spread of illegal and irregular information at the source. At the same time, mainstream app stores such as Huawei, Xiaomi, OPPO, and vivo have further tightened developer access standards and listing criteria, frequently conducting spot checks and retests on applications in the store, successfully intercepting a large number of illegal apps before they entered the store.

To help the public clearly understand the forms of chaos, the official also reported several representative typical violation cases. Some accounts use AI technology to "modify" classic works such as "Romance of the Three Kingdoms" and "Journey to the West," maliciously subvert the images of important characters and fabricate low-quality and sensational plots to attract traffic. Such content has been metaphorically called "digital garbage" by the cyberspace administration department. Some accounts use AI to synthesize false disaster scenes and rescue footage to fabricate false social hotspots or create AI digital people for false promotion and sales to mislead the elderly group.

Some accounts use AI "face swapping" and "voice swapping" technology to generate lowbrow dance videos or kissing and hugging videos by impersonating public figures for marketing and profit. In addition, some platforms have mass-posted strange and terrifying, violent and vulgar content, even creating false scenarios targeting minors such as pregnancy and fighting, spreading "toxic animations" and children's "subversive" videos containing violence, terror, or soft pornography, seriously harming and mocking the image of minors. There are even a few unscrupulous individuals who use AI "management" software to simulate real people operating accounts and generating homogenized copy as online water soldiers. Some intelligent platform covers and conversation contents are suspected to include obscene and pornographic information, and relevant departments have taken strict measures such as warnings, fines, removal, or shutdowns according to law.

The cyberspace administration department stated that in the next step, it will continue to closely monitor new situations and risks brought about by AI-generated and synthesized content, carry out routine prevention of technology abuse, and provide a solid guarantee for building a clean and bright online space.