Amazon’s Purchase-History Profiling Tagged One Customer’s Body Shape Detail

An Amazon customer scrolling through her data learned the company had labeled her as having “flat buttocks,” a detail inferred from what she bought. The post, shared under the pseudonym @fangirlinmegan, ricocheted across social media and pulled back the curtain on just how far purchase-history analytics can go. Behind the awkward phrasing sits a larger machine of NLP, predictive clustering, and “phantom identity” modeling that can translate shopping habits into guesses about bodies. The incident is now landing in a familiar fight over privacy, especially in Europe, where GDPR-style rules can treat this kind of intimate profiling as a line that requires explicit consent.

Profiling purchases, profiling people

If you’ve ever clicked through Amazon’s account settings out of pure curiosity, you know the company keeps more than just your order history. On October 8, 2026, a shopper using the pseudonym @fangirlinmegan posted on Threads that she found a blunt label inside her Amazon profile: the system had inferred she had “flat buttocks,” apparently based on past purchases. The screenshot spread fast, topping 1,000,000 views and sending plenty of people hunting through their own settings.

According to her post, the label appeared under a path like “Your shopping preferences” and then “About you,” where many users expect harmless interest tags. In her case, the tag felt uncomfortably specific, the kind of detail that reads less like personalization and more like a private assessment that escaped into daylight.

How Amazon’s algorithms make these predictions

This kind of inference doesn’t require a tape measure, just correlations. Amazon can apply natural language processing to product titles, descriptions, and reviews, then connect those cues to shopping behavior at scale. Items marketed around “lift,” “shape,” or “push-up,” for example, can become signals that the model treats as proxies for body-related intent.

From there, clustering techniques can do the rest: group shoppers with similar carts, browsing patterns, and reorders, then assign likely attributes to newcomers who match the pattern. That is how a “phantom identity” forms, a profile you didn’t fill out but that still influences what you see, from apparel recommendations to body-focused ads.

Privacy, ethics, and the American regulatory angle

The eerie part is not that personalization exists, it’s the category of what’s being personalized. When an algorithm starts guessing anatomy, it risks turning insecurities into a targeting strategy, nudging shoppers toward corrective products again and again. It is also a reminder that “inferred” data can feel just as sensitive as data you explicitly provided.

In the US, there’s no single federal privacy law that mirrors Europe’s GDPR, but scrutiny is rising. A pattern of intimate profiling could attract attention from regulators like the FTC, especially if consumers argue they were misled about what’s being inferred or how it’s used for marketing.

What trust looks like in the age of inference

The practical question for Big Tech is simple: if a user goes digging in settings and finds something unsettling, does the company have a clear explanation and a clean off switch? Amazon and its peers have spent years teaching us that data makes shopping smoother, but moments like this test whether “smoother” is worth the feeling of being analyzed too closely.

For now, the viral post is less a scandal than a signal. As AI gets better at guessing what we meant, not just what we clicked, the line between convenience and intrusion is going to keep moving, whether companies want it to or not.