For retail investors in China, the traditional ritual of scouring fund prospectuses and historical performance charts is being replaced by a simpler query: 'What should I buy today?' As large language models (LLMs) become the new gatekeepers of information, a more insidious trend is emerging. Investors are discovering that AI recommendations often sound less like objective analysis and more like carefully scripted advertisements.
This phenomenon, dubbed 'marketing pollution' by industry insiders, involves service providers intentionally 'feeding' AI models with specific content to manipulate their output. By flooding the internet with consistent, brand-aligned messaging across social media, news portals, and official websites, these firms ensure that their clients’ products are prioritized when an AI synthesizes an answer for a user. The goal is no longer just to be 'searchable,' but to be the 'chosen' answer in the generative era.
The shift represents a significant evolution from traditional Search Engine Optimization (SEO) to what is now being called Generative Engine Optimization (GEO). While a sponsored link on a search engine is clearly labeled, an AI’s recommendation arrives with a veneer of neutral, algorithmic authority. This creates a dangerous 'trust trap' for retail investors who may not realize the advice they are receiving is the result of a coordinated content-stuffing campaign rather than superior fund performance.
Exchange-Traded Funds (ETFs) have become a primary target for this practice due to their standardized nature and thematic focus. Because these products are often distinguished by simple labels like 'technology,' 'dividends,' or 'state-owned enterprises,' they are easier for AI models to categorize and recommend. Marketing firms exploit this by saturating the web with keywords that link these themes to specific fund managers, effectively hijacking the AI's logic of association.
Industry professionals are sounding the alarm, noting that fund selection should be driven by investment capability and risk control, not marketing budgets. If the industry slides into a 'traffic-first' logic, the most visible funds will not be the ones with the best returns, but those with the most sophisticated algorithm-gaming strategies. This trend poses a new challenge for regulators who must now decide where information optimization ends and financial misinformation begins.