Recently, some netizens found that after Xiaomi released its second product series, Xiaomi Pengcheng (SkyNomad), a large number of search terms and related words on many web pages appeared in association with SkyNomad, such as "SkyNomad car accident", "SkyNomad fire", "SkyNomad owner rights", etc., but Xiaomi Pengcheng's new car was not officially listed for sale, and even just completed the declaration action of the Ministry of Industry and Information Technology on July 10. When ordinary netizens search for "SkyNomad" in the search bar, the search recommendation bar will pop up a large number of derivative keywords with strong tendencies and full of negative orientation. Industry professionals explain that the search engine recommendation mechanism relies on user search behavior clustering generation: if the same kind of negative words are repeatedly searched in a short period of time, the system will automatically aggregate and display the relevant entries. For this model that has not yet been introduced to the market, manufacturing large-scale false retrieval data from scratch can basically determine that it is a malicious operation planned by humans. At present, there are two main types of "data poisoning" methods for polluting AI algorithms in the industry: first, manipulating a large number of accounts to focus on high-frequency retrieval of fabricated negative keywords and inputting abnormal retrieval traffic to the platform; second, publishing relevant content bound to negative terms in batches for a short time and then deleting them in batches to pollute the underlying recommendation model of the platform. In addition, a large number of actual videos of traffic accidents have emerged on major short video platforms. The publisher deliberately moved flowers and trees, and insured unrelated car accidents on the Xiaomi Pengcheng SkyNomad, which has not yet been released. The poster falsely said that "the newly acquired Xiaomi SkyNomad has another traffic accident", and at the same time, the spread of negative labels such as models and accidents was superimposed. Although many short video platforms have labeled AI-generated prompts for such content, reminding viewers to distinguish between authenticity and falsity, the continuous spread of large quantities of false material has caused irreversible pollution to the information base of AI question-answering models. In some of the existing AI-generated responses, the accident-related vehicles were mistakenly identified as Xiaomi Pencheng SkyNomad, which has not yet been listed. According to the analysis of industry practitioners, this disinformation model is essentially different from the traditional offline and online publishing of paid black manuscripts: the fake content produced by AI has three prominent characteristics: low production cost, mass distribution, and difficult to trace the source. Once used as a commercial public opinion attack tool, it can form a wave of disinformation in a short period of time, quickly distorting public perception and causing serious damage to the entire network information environment. (Compiler/AutoHome graduated).