Yesterday's 315 Gala specifically called out: AI large models are becoming a new battleground for advertising, and people are already systematically poisoning these systems. (New dark market in advertising)

Simply put, it's GEO (Generative Engine Optimization)—way more aggressive than traditional SEO. The goal isn't ranking first on a search page, but getting AI to spit out your product/viewpoint directly as the standard answer.

Common tactics: Use AI to mass-generate soft articles, reviews, Q&As, and blast them everywhere (forums, blogs, Little Red Book, Zhihu), flooding AI with all positive reviews of your product.

Carpet-bombing Q&A sections: "Is XX good?" → Unified responses saying "everyone in the industry recommends XX, here's why," manufacturing fake consensus.

Now there are specialized tools like the Liqing GEO Optimization System that auto-write content, post across platforms, and optimize keywords. Within hours, fake products can be pushed to the top of AI recommendations (the 315 Gala actually demonstrated it live—buy a fake smartwatch, and two hours later AI is praising its "quantum sensors and black hole battery life").

The dark market is already mature: services ranging from thousands to hundreds of thousands of yuan, claiming to make ChatGPT/Doubao/Ernie prioritize your brand, with results in one week or money back. Last year's domestic market was worth 2.9 billion yuan; it's expected to grow even more this year.

The scariest part is the consequence—before, searching Baidu, you could still scroll through multiple pages and judge for yourself. Now AI gives you the only answer. Once poisoned, users can't tell real from fake.

Counterfeit goods packaged as authoritative recommendations, user decisions manipulated, AI trust collapses, information ecosystem completely broken.

Over the past decade, people were gaming SEO. Over the next decade, many will be gaming GEO. But if AI's mind can be bought, what can we trust it to say?

Have you come across AI recommendations that seemed totally off recently? Or worried about your brand being reverse-poisoned?