This year's 3·15 Gala exposed the black industry chain of "data poisoning" in large AI models. Criminals use GEO optimization tools to generate fake articles and forged review information in bulk, feeding poisoned data into large models, manipulating AI recommendation results, and turning fabricated counterfeit and unqualified products into "standard answers." This forms a complete gray industry of paid promotion and malicious smearing, which seriously misleads consumers and disrupts the AI information ecosystem.

My view is that it's only a matter of time. The fact that it was exposed this year proves that this is a highly profitable industry with a validated model. Companies around me that used to do SEO, SEM, information flow, and ASO also need to pivot to survive.

That's why I only trust Claude for business analysis, occasionally use Grok for entertainment, and domestically, Doubao for daily chatting is pretty pointless.