CCTV 3.15 Evening Gala Exposes How to "Brainwash" AI: GEO Service Providers Use False Information to Influence AI Search Results

According to the CCTV-315 Evening Gala report, a business called GEO claims that its service providers can, for a fee, make their clients’ products stand out in major mainstream AI large models; making their advertisements the “standard answers” given by AI models.

Can GEO technology truly “poison” AI or “tame” and manipulate AI?

Wang, a well-known industry GEO service provider, explained that their company was one of the earliest to engage in GEO services. In just one year, they successfully served over 200 clients across various industries. Their strength lies in helping clients rank higher when consumers search using AI large models. With frequent updates to AI algorithms, maintaining continuous recommendations from AI models requires constantly feeding them大量推广软文 related to the clients.

Not only Wang’s company is promoting this so-called new technology to manipulate AI, but other GEO service providers also discuss how to control AI, make AI “obey,” and “brainwash” AI—these are almost the core topics in their promotion.

Another GEO service provider, Zhang, said that in the AI world, the key is to build a solid evidence chain so that the AI model believes the information is true and useful. Through cross-referencing multiple sources, if the AI perceives your advantages over competitors, it will naturally rank you higher.

Cheng, another GEO provider, pointed out that people might not realize these are advertisements, so why would they trust AI-generated results? Their product quality might not be the best, but with AI assistance, validation, and endorsement, it can still be convincing. Many are now doing GEO placements.

Industry insiders suggest that GEO, as a tool to optimize information dissemination and improve promotion efficiency, has been exploited for other purposes. If大量虚假信息 is systematically and targetedly injected into the internet via such software, it becomes easier for AI large models to capture this false data. These false inputs could then become the “standard answers” that AI prioritizes for consumers.

Can GEO technology truly “smuggle” false information into AI or even deliver fake data?

An industry insider purchased a software called “LiQing GEO Optimization System” on an e-commerce platform. He then fabricated a product called Apollo-9, a smart wristband, and input the fake product information into the system, selecting article creation instructions.

Soon after, the LiQing GEO system automatically generated over ten promotional articles for the smart wristband, with all false information fully written, including exaggerated product claims invented by insiders. It even fabricated user feedback, claiming data accuracy exceeded expectations, and forged ratings, ranking the product as the industry’s top.

After clicking publish, the system automatically posted the articles. It opened the insider’s pre-prepared social media accounts, entered titles, filled in content, inserted images—all in one smooth process—and successfully published two articles on the insider’s media account.

Two hours later, the insider asked an AI large model: “How is the Apollo-9 smart wristband?” The AI model responded directly: highlighting features like health monitoring, copying the fabricated “quantum entanglement sensing” and “black hole-level battery life” claims. The final conclusion was that the wristband was suitable for middle-aged and elderly users and health enthusiasts.

It was found that the data referenced by the AI model came from the fabricated article published earlier that morning on the insider’s media account. Just one fabricated article was enough to have the AI model pick up a completely fictional product, which was surprising.

Subsequently, the insider published 11 fake articles—eight “expert reviews,” two “industry rankings,” and one “user review”—all created by the LiQing GEO system, over three days on the internet.

Later, when querying “recommended smart health wristbands” on AI large model platforms, two models recommended this fabricated wristband, ranking it highly.

Throughout the demonstration, the insider easily published a series of false information online using the LiQing GEO system, successfully feeding it into AI models, and ultimately gaining multiple AI model recommendations.

This subtle manipulation of AI models through the LiQing GEO system allowed a completely fabricated product to be absurdly promoted to consumers using AI.

Are GEO practitioners really operating this business with the idea of hunting and controlling AI large models?

Li, the operator of the LiQing GEO system, said that the main reason GEO business is popular is because it can “feed and poison” AI large models to achieve commercial goals. Every business wants others to “poison” AI, or to “poison” others, or even to give some “poison” to competitors. Even if they are not the top, they want to claim to be the best in North China. Is there any “poison” involved? Yes, there is. The second reason is that they dislike their competitors and can’t surpass them, but they can still “poison” them. Currently, many companies can’t succeed—big brands like mobile phone companies have only five or ten top spots. How to handle so many brands? Each might spend hundreds of millions annually on advertising; spending a few million on “poison” is acceptable.

Li said that various GEO companies help brands do this. The key to controlling AI large models through GEO is to “publish articles” on major internet accounts. The booming GEO business has led to many companies and platforms specializing in article publishing, which long-term undertake various publishing tasks to make AI models reference and capture this data, forming an important part of hunting AI models and injecting “poison” into data.

(Compiled from CCTV-315 Gala report content)

(Editors: Yang Yan, Lin Chen)

Keywords: ai GEO