Imagine you want to post a product review on a platform like Xiaohongshu—let’s say a pair of wireless earbuds you’ve never tried. You’ve seen a few buzzwords floating around: “adaptive noise canceling,” “transparency mode,” “comfortable fit.” So you open an AI writing tool, type in the product name, dump in those keywords, hit generate. Seconds later, out pops a full-fledged “real experience” note: “Girls, I’ve used these for three months—the noise canceling on the subway saved my life, my commute feels like a spa…” You slap your name on it and post. Is that your creative work? Or is it just pure fiction?
The answer isn’t as blurry as you might think. Recently, the Hangzhou Intermediate People’s Court in Zhejiang dropped the first major case in China involving AI-generated fake reviews—and it’s a wake-up call for anyone dipping a toe into AI content creation. The plaintiff was Xiaohongshu (Little Red Book), a platform that lives and breathes authentic user recommendations. The defendants? A company that sold an AI tool specifically designed to churn out those very same “种草” notes. The tool had modules labeled “Xiaohongshu Product Reviews,” “Xiaohongshu Travel Guides,” “Xiaohongshu Titles.” You get the picture—it was built to game one platform’s ecosystem.
The court didn’t buy the usual “tech is neutral” defense. Think of it this way: a knife seller doesn’t get blamed when someone uses their knife to rob a store—but if they sell a knife with a custom grip and a step-by-step manual for picking locks? That’s a different story. The court laid out a four-step test to decide whether an AI tool crosses the line into unfair competition. First, is it a generative AI service? Not a simple template-filler, but one that literally creates new text from scratch. Check. Second, is it targeted at a specific platform? The modules literally had the platform’s name in them—check. Third, does it harm the platform’s core value? Xiaohongshu’s value is authenticity—mass AI-generated fake reviews destroy that. Fourth, does the tool’s business model depend on circumventing the platform’s rules? The company sold memberships to let users pump out fake reviews—double check.
Here’s the kicker: the court didn’t just slap the defendant—it gave us a clear ruling on where the line is. If you’re building a generic AI writer that helps people with emails, essays, or even casual social posts, you’re probably safe. But if you design a tool to target one platform’s content rules and profit by helping users bypass them, you’re on thin ice. Remember, knives aren’t the problem—it’s who you sharpened them for and what you told them to cut. So next time you think about using AI to fake a review, ask yourself: am I being creative, or am I just being a digital pickpocket? The law is watching, and it’s got four clear lines.