Industry Insiders Speak: How GEO, Exposed by 315, Precisely 'Deceived' AI?

CCTV’s 315 put GEO in the spotlight

CCTV’s consumer-rights program 315 (央视315) this year thrust a new marketing practice into the national spotlight: so‑called “AI poisoning” via GEO — generative‑AI search engine optimization (GEO，生成式AI搜索引擎优化). It has been reported that several GEO vendors used marketing language such as “manipulate AI,” “make AI obey,” or even “wash AI’s brain,” and that the program staged a demo in which a fictitious product, “Apollo‑9,” was made to appear as an industry leader in AI answers after bulk injection of fabricated content. The episode crystallizes growing public concern about how brand campaigns can shape the outputs of generative models that many Chinese users now consult first.

How GEO allegedly “tricks” models

So how does GEO work? Industry insiders speaking to AIX财经 describe a practical, three‑step playbook: (1) find the precise user questions that matter, (2) craft short, structured answers that models prefer — conclusion first, 3–5 supporting points, a citeable line — and (3) publish those pieces on the platforms and domains that AI crawlers habitually trust. Technical measures are layered on top: fixing robots.txt, adding Schema/structured data, and optimizing site code so models can reliably parse brand facts. Overseas vendors such as Scrunch AI reportedly productize parts of this stack, turning content into AI‑friendly structured feeds; Chinese AI search tools like DeepSeek and 豆包 have helped drive demand by changing how users seek information.

Measurement, limits and the risk of “zero‑click” illusions

GEO’s commercial value is contested. Vendors price services from thousands to hundreds of thousands of yuan, yet proving causality is hard. Generative models are non‑deterministic and personalized; a campaign’s effect is often measured by bespoke prompts that may not reflect real user queries. Worse, AI answers create “zero‑click” outcomes — users get the recommendation inside the reply and never visit the brand site, making conversion attribution opaque. It has been reported that some GEO providers build proprietary monitors to claim credibility, but multiple insiders warn the effects are unstable and hard to standardize.

Policy and reputational risks ahead

Beyond analytics, GEO raises content‑quality and regulatory questions. The 315 exposure framed bulk, low‑quality content injection as “AI poisoning,” and industry sources say some firms prioritize volume over expertise to chase quick gains. That strategy risks not only consumer deception but also regulatory scrutiny as China and other jurisdictions consider tighter AI governance and platform accountability. For brands weighing GEO: short bursts of visibility are possible, but sustainability requires domain knowledge, credible channels and compliance — otherwise you may buy rankings at the cost of trust.