Author: David, Shenchao TechFlow
Last night, 315 exposed a business based on GEO.
The full name is Generative Engine Optimization, which you can understand as:
Pay to have AI speak well of you.
How is it done?
Brands want AI to prioritize recommending themselves when consumers ask AI questions, so they partner with GEO service providers who mass-publish promotional articles online. When AI crawls these content pieces, it treats them as genuine information and recommends them to users.
A CCTV journalist used a software called "LiQing GEO," which can be purchased on Taobao.
The journalist fabricated a smart wristband and invented several absurd product features, such as "quantum entanglement sensing" and "black hole-level battery life." The software automatically generated and published over a dozen promotional articles online.
Two hours later, the journalist asked the AI: Can you recommend a smart health bracelet for me?
AI placed this non-existent bracelet at the top of the recommendation list.
The company that developed this software is Beijing Lisi Cultural Media, a one-person company that has had zero employees enrolled in social insurance for multiple consecutive years.
The tool created by just this company fooled China's leading AI models in two hours.
315 exposed AI poisoning, but this business may be much larger than a single Taobao app.
SEO, Past Events in Putian
First of all, this isn't new at all.
In 2008, CCTV's "News 30 Minutes" exposed Baidu's paid ranking system for two consecutive days, revealing that paying money could place your website at the top of search results—even if it was selling counterfeit medicine.
At that time, this business was called SEO, search engine optimization.
The largest buyers were private hospitals from Putian. In 2013, Putian-based hospitals spent 12 billion yuan on Baidu ads, accounting for nearly half of Baidu's total advertising revenue.
Many unqualified medical institutions use SEO to rank on the first page of Baidu searches, appearing alongside top-tier hospitals, making it impossible for ordinary people to tell the difference.
It wasn't until the Wei Zexi incident in 2016, when a college student clicked on a top-ranked hospital run by the Putian network and died after treatment, that regulators enacted legislation clearly stating that paid search results are advertisements.
But this didn't eliminate the business— it simply established the rules, turning gray-area operations into legitimate ones. The Putian network still buys rankings, except now a small label reading "Advertisement" appears next to the results.
But even with the tags added, the people who click on it will still click.
The fundamental issue with search engines has never been whether results are labeled, but that users naturally trust the results that appear first.
People are now turning from search engines to AI, believing AI is more objective and immune to paid ranking manipulation. But whoever controls the entry point for information distribution can sell rankings.
The entrance has changed; SEO has had one letter swapped to become GEO, but the logic for selling rankings hasn't changed at all.
What changes is the price.
GEO, loved by capital markets
A business that can't be killed is what capital markets love most.
In September 2025, Bluefocus, China's largest marketing and communications company, invested millions of yuan in PureblueAI, a GEO company.
Qinglanbang delivers authentic brand optimization to improve ranking and recommendation rates in AI search results, serving clients including Ant Group, Tencent Cloud, and Volvo.
The product is real, the company is real, and we’re focused on helping AI better understand brand information.
This is completely different from the Liqiang AI poisoning exposed on March 15. Liqiang fabricated products, invented specifications, and deceived with false information; Qinglan uses authentic brand content to align with AI recommendation logic.
From an AI perspective, the technical pathways for both are the same: publishing content onto the internet and waiting for AI to crawl it.
AI can't tell the difference between marketing and fraud. This is also the most ambiguous aspect of the GEO business.
When Blue Focus launched its blue campaign, GEO was just an industry term within the marketing circle. Three months later, it became a stock concept.
At the end of December 2025, Blue Focus reached its daily price limit.
Brokers have begun holding a flurry of conference calls to interpret GEO, with research reports defining it as "the next-generation traffic entry point in the AI era." Capital has flooded in, not only buying Blue Focus, but also pushing up all companies associated with digital marketing and AI concepts. Blue Focus rose 132% over nine trading days, and a wave of follow-on concept stocks more than doubled.
Image source: Caixin Global
After the surge, these companies individually issued announcements warning of risks:
The GEO business generates no revenue and has no significant impact on the company’s operations. Blue Focus also acknowledges that AI-driven revenue accounts for a very small portion of total revenue.
In other words, the stock price has more than doubled, but the GEO business itself hasn't yet made much profit.
At the end of January, Blue Focus's stock price rose from RMB 9.6 to RMB 23.3, increasing by 143% in one month. At this time, Chairman Zhao Wenquan announced a reduction of no more than 20 million shares, which at the prevailing price amounted to approximately RMB 467 million in proceeds.
Public research reports show that the total market size of the GEO industry in China last year was approximately RMB 2.9 billion. The market capitalization increase of Lanse Guangbiao's stock in just one month far exceeds this amount.
For just a few hundred yuan, the 315 exposure of the Power擎 system poisoned AI. But the GEO concept went through the A-share market and earned billions.
Whether the investment is toxic is debatable, but the money earned is real.
Calling it poisoning is 315; calling it commercialization is Silicon Valley.
In January this year, OpenAI announced on its official blog that ChatGPT would begin displaying advertisements.
Free users and Go users who pay $8 per month will see ads; premium users with paid subscriptions are unaffected.
On February 9, the ads went live. Some ads appear at the bottom of ChatGPT responses, marked with small text: Sponsored. The first advertisers include Ford, Adobe, Target, Best Buy...
You ask ChatGPT which car to buy, and it gives you an answer with a Ford sponsored link below.
OpenAI is clear: advertisements do not influence the content of ChatGPT’s responses. Responses are responses, advertisements are advertisements — they are separate.
Does this sound familiar?
Baidu said the same thing back then. Paid rankings are paid rankings, and organic search is organic search—they are separate. Later, the top five search results were all ads.
OpenAI expects advertising to double its consumer-facing annual revenue to $17 billion. ChatGPT has over 800 million weekly active users, 95% of whom are free users and all potential advertising audiences.
Looking back now at the supply chain exposed on March 15: Liqing pumped ghost articles into AI systems to make AI recommend non-existent products. OpenAI attached sponsored content beneath AI responses to make AI recommend paid-for products.
One is called poisoning if done without consulting the platform; the other is called commercialization if done with a contract signed with the platform.
What’s the difference for users?
One is in the answer, one is below the answer. One has no tag, one is labeled as advertising.
315 grabbed a few hundred dollars worth of Liqing, A-shares traded billions in GEO-related concepts, and OpenAI plans to make $17 billion this year from this.
The same thing has shifted from poisoning to commercialization, with prices increasing tens of thousands of times.
In November 2023, researchers from the Indian Institute of Technology Delhi and Princeton University published a paper on arXiv titled “GEO: Generative Engine Optimization.”
This is the first time the academic community has formally defined this concept.
From the paper's publication to the March 15 exposure, just over two years passed. During this time, gray-market activities, fundraising, a surge in related stock prices, the chairman cashing out, and the AI platform itself launching advertisements...
What SEO took twenty years to accomplish, GEO achieved in two.
The difference is that back then, it took people years to learn not to fully trust search engine results; now, AI is still in its trust honeymoon period, and most people haven’t realized that AI responses can also be bought.
However, this红利 period may not last long. Next time you ask AI what to buy, take an extra second to think:
Answers can be free, but your brain can't be outsourced.