The "Blue Book on Internet Advertising Governance in the AIGC Era" was released, advocating "using AI to govern AI."

On October 24, during the 33rd China International Advertising Festival, the "Blue Book on Internet Advertising Governance in the AIGC Era" was officially released. While AIGC technology brings efficiency gains, the other side is escalating risk: black and gray industries use AI to mass-generate highly realistic materials and forge authoritative endorsements, making manual review alone increasingly inadequate. Facing these new risks, the blue book provides governance methods spanning the entire advertising delivery process: classifying risks into four categoriescontent violations, technical confrontation, transaction links, and new scenariosand offering recommendations from eight aspects. For example, the blue book presents the practical approach of "using AI to govern AI": the first step in governance is to see the risks clearly. Multimodal large models are used to simultaneously view images, read text, and listen to audio, enabling them not only to understand surface meanings but also to connect context and determine whether violations exist. Xiao Tao, head of content security strategy for business security at Ocean Engine, introduced that relying on the intelligent review system, 90% of materials are reviewed within 10 minutes at the fastest, with efficiency improved by 75%. In addition, AI application platforms also incorporate user feedback into governance, with special protections for the elderly and minors. The elderly can provide evidence with "zero threshold" and reach human customer service directly, while parents use standardized templates to report problems, with some issues resolved the same day. After standards are unified, platforms maintain routine risk monitoring and inspections, conducting advance research on risk trends around high-incidence problems; for new variants that have changed appearance, they quickly identify and promptly update interception strategies, promoting the advancement of identification and handling; at the same time, they continuously iterate identification models and review rules, strictly handling entities involved in malicious forgery and repeated violations until they are suspended and removed, and reporting suspected illegal and criminal activities to relevant authorities in accordance with the law or cooperating with investigations.