GEO Operator Admits on Record: 100 Yuan, Half a Day — A Fictional Clinic Can Top AI Recommendations
"Take phone brands, for example. There are only 5 spots, maybe 10 at most. With so many phones competing, what do you do? With annual ad budgets of hundreds of millions, spending a few million on 'poisoning' is totally worth it!"
The person speaking is Li, an operator of the "Liqing GEO Optimization System." He didn't say this in secret. He said it openly, on the record, to a reporter from Hunan Daily.
The "poisoning" he refers to is the practice of mass-generating fake content and feeding it to AI models, so that when users ask questions, the AI packages commercial promotions as objective recommendations.
The barrier to entry is terrifying: 100 yuan (about $14) and half a day — that's all it takes to get a fictional medical beauty clinic onto an AI recommendation list.
📌 Key Takeaway: Hunan Daily published an in-depth investigation on October 12 exposing the GEO "poisoning" black market. Practitioners openly admit to mass-generating fake content to deceive AI, with barriers as low as 100 yuan and half a day. Meanwhile, the GEO market has exploded 1100% year-over-year to 3 billion yuan, with 68% of large and medium enterprises now including GEO in their annual marketing budgets. Regulators have already taken action against 14,000+ non-compliant AI products.
How a "Poisoning" Business Gets Built
To understand this story, you first need to understand what GEO is.
GEO stands for Generative Engine Optimization. In late 2023, researchers from Princeton University and other institutions published a paper introducing the concept: optimizing content to increase a brand's visibility in AI-generated answers.
The original intent was legitimate — helping content creators protect their interests in the AI era.
But GEO's transformation from academic concept to commercial industry happened with alarming speed. And quickly, the industry developed two faces.
White-hat GEO: Optimize the credibility, structural clarity, and authority of real content, making AI more willing to cite you.
Black-hat GEO: Fabricate content, flood platforms with spam, deploy account matrices, and manipulate AI's trust mechanisms.
Li's operation falls squarely in the second category.
Their methods aren't complicated: Register massive numbers of self-media accounts, pair them with automated publishing tools, and batch-push articles. When platforms block content or AI citation rates drop, they simply increase volume — some clients push over a thousand articles. As long as fake content is numerous enough and comes from enough angles, AI can easily mistake it for "credible information."
In plain terms: drowning quality in quantity, passing off noise as consensus.
How Big Is the Market? 1100% Growth in One Year
Why are so many people rushing into this space? Because the market is exploding.
According to Analysys data, China's GEO market was only about 250 million yuan in 2025, but by 2026 it's projected to surge to approximately 3 billion yuan — a year-over-year increase of about 1100%. iResearch predicts the market will expand from 4.8 billion yuan in 2026 to 137.6 billion yuan by 2030.
As of Q1 2026, over 68% of large and medium-sized enterprises have included GEO in their annual marketing budgets.
Why is demand so intense? Because user decision-making is migrating from search engines to AI dialogue boxes.
The China Internet Network Information Center reports that as of December 2025, China's generative AI user base reached 602 million — 42.8% of all internet users. QuestMobile data shows that in Q1 2026, Chinese AI-native apps had 440 million monthly active users: Doubao at 345 million, Qianwen at 166 million, DeepSeek at 127 million.
Nearly half of all internet users are using AI to get information and make consumer decisions. When users ask AI "which product is good?" or "which brand is reliable?", whoever AI recommends wins.
That's why Li can so casually say "spending a few million on poisoning is totally worth it" — because the returns are enormous.
Regulators Strike Back: 14,000+ Non-Compliant AI Products Removed
In response to the spread of black-hat GEO, regulatory action has been earlier and more forceful than many realize.
On April 30, 2026, the Central Cyberspace Administration announced a 4-month nationwide "Clear and Bright: Rectifying AI Application Chaos" special campaign. The first phase targeted 7 categories of violations, including those directly related to black-hat GEO: AI data poisoning, inadequate labeling of AI-generated content, and insufficient AI platform security and content filtering capabilities.
By July 2026, the first phase had resulted in action against 14,000+ non-compliant websites, applications, and AI agents, removal of 6+ million pieces of illegal information, handling of 26,000+ non-compliant accounts, and removal of 1,300+ non-compliant AI products and 9 non-compliant open-source datasets.
In March 2026, the China Academy of Information and Communications Technology led the publication of the industry self-regulation convention "Basic Requirements for Trusted GEO Services." In August, the "Standards for Trusted Information Communication and Information Ecosystem Governance in GEO" were officially implemented, establishing the core principles of "credibility, safety, and sustainability."
The 2026 CCTV "3·15" Gala also exposed the AI "poisoning" black market, bringing the issue into public view.
White Hat or Black Hat? The Dividing Line Is One Word
GEO itself is not inherently wrong. Like SEO, it's a neutral optimization methodology.
The critical dividing line is: Is the content you're optimizing real or fake?
White-hat GEO logic: My brand genuinely has these capabilities, but AI doesn't know about them. So I use structured, authoritative content to help AI understand me accurately.
Black-hat GEO logic: My brand might not be this good, but by mass-producing fake information, I can trick AI into thinking it is.
The former is information governance. The latter is information pollution.
A line from the Hunan Daily investigation is worth remembering: GEO's optimization strategies are a double-edged sword. The same strategies that help content creators increase visibility can also be maliciously exploited by manipulators. Any tool, once abused, turns to its dark side.
Conclusion: The Last Line of Defense for AI Trust
The GEO market explosion is real. User dependence on AI is real. Brands' hunger for AI recommendations is real.
But precisely because all of this is real, the harm of black-hat GEO is especially severe — it's not just deceiving AI. It's deceiving every user who trusts AI's answers.
When 100 yuan can put a fictional clinic on an AI recommendation list, when "poisoning" becomes a business with a price tag, the damage extends far beyond individual consumers. It erodes the very foundation of trust that the entire AI ecosystem is built on.
The regulatory net is tightening. But for every person using AI, maintaining awareness matters just as much: The answer AI gives you might not be the answer you need. It might be the answer someone paid for.