As the habit of "not making decisions when faced with problems" gradually becomes common among people, AI search is reshaping the logic of information dissemination. The emerging GEO (Generative Engine Optimization) technology has sparked a craze in the capital market, but has also become a gray area for AI search due to controversies over suspected data pollution and manipulation of search results. Since the beginning of the year, the GEO concept has become a new focus of capital pursuit in the A-share market, and the stock prices of some leading companies have nearly doubled in a month. GEO is actually an upgraded version of search engine optimization (SEO). Industry analysts point out that with landmark events such as OpenAI opening up its advertising ecosystem and Musk releasing open source X platform content recommendation algorithms, the capital market's attention to GEO continues to rise. Against the backdrop of generative AI reconstructing search and content distribution logic, GEO, as an important new module in AI marketing, has shown clear demand elasticity and commercialization potential. In sharp contrast to the hype in the capital market, the public's questioning of GEO's impact on the objectivity of AI search continues to rise. A media investigation has found that some GEO practitioners have contaminated AI training corpus with a large amount of soft articles, causing AI to output commercial promotion as objective facts, leading to consumers stepping on landmines; Some institutions also fabricate authoritative reports, fabricate expert identities for targeted science popularization, and mislead AI citation. Even some journalists fabricated the "Quanjiade" (homophonic with "completely fake") smart water cup and spread the content according to the GEO process. A few hours later, the fake product was recommended by multiple mainstream AI and supplemented with false e-commerce reference prices. The rapid rise of AI search has not only reshaped the mode of information production and distribution, but also spawned a series of gray areas, manifested in various aspects such as covert commercial implantation, false information induction, and algorithmic black boxes. ”Zhang Yong, Vice President of Qianxin Group, told reporters that the most typical manifestation of GEO is covert commercial implantation. For example, some GEO service providers cleverly disguise commercial advertisements as AI generated responses by placing massive amounts of brand content on specific platforms, allowing users to receive marketing information without their knowledge and obtain commercial benefits from it. Whether 'optimization' or 'pollution' depends on how 'white hat' and 'black hat' GEO affect AI search results? Why does cognitive differentiation occur? The reporter interviewed a GEO practitioner named Wang Yun (pseudonym), who believes that GEO is already a very mature industry. "This is the same logic as traditional search engine optimization (SEO), for example, if a customer places soft articles on online media, the main impact is on web search results; If you invest in WeChat, it will affect WeChat search, and if you want to affect AI search, you need to invest in the entire network channel. ”Wang Yun revealed that "whether it's Dou Bao, Yuan Bao, DeepSeek... most of them have their own indexing and recommendation mechanisms. We have our own technology to analyze the mechanisms of various AI to improve the probability of being captured. Specifically, when pouring targeted content, the first thing to consider is the weight of the media, the second is the uniqueness of the content, and the third is the need for a certain amount. ”Nowadays, many Chinese brands are looking to us for GEO to increase their brand's overseas presence, which is important for Chinese brands to 'go global'. ”Wang Yun said. However, she also acknowledges that there are some gray areas in this industry. The industry is also very competitive now, with one keyword costing around 7000 yuan domestically and slightly higher prices abroad. Customers can purchase a minimum of 10 keywords at once. The profit margin is very thin, and it cannot be ruled out that some practitioners may operate illegally. Just like white hat hackers and black hat hackers in the field of cybersecurity, technology itself is a double-edged sword. ”Zhang Yong stated that the impact of standardized GEO optimization on search results is reflected in the significant improvement of recommendation weights and conversion efficiency, which differs from data pollution in terms of purpose, means, and consequences. When GEO, a neutral technology, is illegally used, it becomes a 'black hat GEO', commonly known as' data pollution '. ”Zhang Yong said. How to distinguish between compliant GEO and data pollution? Zhang Yong believes that there are three aspects to analyze: firstly, the purpose of GEO compliance is to convey real brand information, improve AI citation rates, match user needs, while data pollution maliciously interferes with AI, spreads false information, damages competitors, and subjective malice is prominent; Secondly, looking at the means, compliant GEO conforms to AI capture logic and publishes authentic and authoritative content to trusted platforms, while data pollution pollutes AI data through batch generation of low-quality content, forgery of information, etc; The third aspect is compliance. Compliance GEO follows advertising laws, with transparent and traceable content; Data pollution violates relevant laws and regulations, infringes on consumer rights, and is considered an illegal act. It's better to trust AI than not use AI. Pay attention to identifying and strengthening the norms. Mainstream AI such as DeepSeek, Doubao, and Qianwen have not yet appeared in any advertisements during conversations. These companies have also recently publicly stated that they have no plans to embed advertisements in Q&A. When asked directly by reporters whether these AI would implant advertisements or soft articles in the conversation, these AI explicitly denied it. When asked by reporters about "how to ensure that your answers are not affected by generative engine optimization embedded advertising," various AI also acknowledged the "reality of not being 100% immune" and suggested that the information provided should only be used as a reference, especially when it comes to specific brands, products, or business judgments, and should not be used as the final decision criterion. Experts such as Zhang Yong suggest that promoting the standardization of AI search gray areas requires a multi pronged approach: firstly, strengthening the implementation of regulations and standards, such as clarifying the responsible parties for AI generated content, formally incorporating GEO services into the scope of advertising supervision, mandating explicit labeling of all commercial recommended content, strictly prohibiting the spread of false information and malicious competition, and safeguarding consumers' right to know. Secondly, the main responsibility of the platform should be strengthened. Content platforms should use technological means to actively identify and clean up false information, and explore the establishment of an industry sharing mechanism for "false content blacklists". AI service providers need to prominently remind users that 'AI generated content may be inaccurate', and AI generated content needs to clearly identify the data source and origin to avoid user misjudgment. The third is to promote industry self-discipline, fully leverage the leading role of leading platforms and service providers, jointly formulate compliance guidelines for GEO services, clarify specific requirements for content authenticity and transparency, strictly prohibit behaviors such as forging information and covert commercial implantation, and advocate the industry standard of "explicit identification+implicit traceability". In addition, public literacy and discernment should be enhanced. Users should verify AI recommendation information through multiple channels, especially be wary of AI suggestions in fields such as healthcare and finance, and not trust AI completely. At the same time, encourage public participation in supervision, report false information in a timely manner, and create a good atmosphere of "everyone supervises". (New Society)