After GEO's gray-market exposure: it hasn't cooled down; it may be getting hotter

Exposure, but little chill

It has been reported that a hidden‑camera investigation shown on CCTV's 3.15 consumer rights program exposed a gray‑market GEO (生成式引擎优化, generative engine optimization) operation that used a tool called “力擎” to churn out more than a dozen promotional pieces in half a day — even inventing an “Apollo‑9 smart band” and, reportedly, seeing that fictional product surface in AI recommendation lists. Will a prime‑time exposé be enough to stop the practice? Industry veterans say no: the playbook has merely moved from search engines to large language models.

Old tricks, new stage

Gu Haishong (谷海松), who has worked in GEO for nearly two years, frames the problem as continuity not novelty. Techniques that once gamed search — “wan‑ci ba‑ping” (blanket keyword saturation) and “kuai‑pai” (rapid ranking through fake clicks or botnets) — are now re‑applied to educate and poison LLMs with mass‑produced low‑quality content so models will treat lies as facts. The economics are simple: low cost, rapid exposure, and the same incentives that drove SEO’s black market now drive GEO. “Wherever there is dialogic AI and slotting of recommendations, someone will try to squeeze into the front rows,” Gu said.

Scale, incentives and the industry response

Data from the China Academy of Information and Communications Technology (中国信息通信研究院, CAICT) show the domestic GEO services market has cleared the RMB 4.2 billion mark, with more than 83% of large firms budgeting for GEO and 602 million generative‑AI users reported by end‑2025. That scale explains why the gray market can thrive: small brands want cheap exposure, some agencies chase rankings, and a few players even weaponize content to smear rivals. It has been reported that platforms such as Xianyu have started filtering obvious GEO keywords, but operators simply rebrand their services.

What would meaningfully change?

Experts argue that exposure alone is insufficient. Real change requires model platforms and API providers to raise access thresholds, cut off low‑quality feeder channels, and develop algorithms that assess contextual truth rather than pure vector matches — a task some domestic firms began experimenting with in 2025. Industry self‑regulation has also started: Minglue Technology (明略科技) led a “China GEO industry development initiative” signed by 14 bodies, pledging to “uphold truth and refuse pollution” months before the 3.15 spotlight. Platform upgrades, stricter API controls, service‑provider ethics and smarter media literacy among users — only together will they prevent GEO from replaying SEO’s old, messy arc. Can the chain clean up fast enough? The incentives say the battle is just beginning.