Somewhere in the Amazon app there is a page that reads like a stranger’s notes about you, and this week a lot of people found it at the same time. One woman posted that Amazon had concluded she “has flat buttocks.” Another found a line under Home & Family stating that they had “never known anyone who has lost a leg in an accident.” Someone else was informed they are too short to reach into a 40-inch box.
The page is called About You. It has existed since May. Almost nobody knew it was there until a post on Threads went around, and now the screenshots are everywhere.
The funny part is genuinely funny. The useful part is that this is the first time Amazon has shown shoppers the actual sentences its personalization model has written about them, and unlike almost every other data page a big platform offers, several of those sentences can be deleted.
Find yours in under a minute
- In the app: tap the Me tab, tap your name, then tap About You
- On desktop: hover over your name above Account & Lists at the top right, click your name or Who’s shopping?, then click About You
- Shortcut: ask Alexa for Shopping “what do you know about me?”
- Then: each line can be corrected or removed individually. There is no single switch that stops the guessing
What the page actually is
About You went live on May 13, 2026, alongside Alexa for Shopping, after being trailed earlier in the year. Amazon’s framing is transparency: rather than a recommendation engine working invisibly in the background, here is the profile driving it, and here is your chance to fix it.
That framing is fair as far as it goes. Amazon has been using machine learning to recommend products for well over a decade, and the company’s own description of the ambition has not changed much: “We aim to make every experience feel like it was designed especially for you, whether you are trying to find the perfect outfit, a book you’ll love, or your next dinner idea.”
What is new is not the inference. It is the sentence. Previously the model concluded you probably buy non-dairy milk and quietly showed you oat milk. Now it writes “prefers non-dairy alternatives” on a page with your name at the top, and the difference in how that feels is the entire story.
What feeds it
Amazon names its inputs, which is more than most platforms do. Five streams go in.
Notice what is doing the work in those viral examples. Nobody told Amazon anything about their body. A shopper bought butt-scrunch leggings, and the model reasoned backwards from the product to a physical characteristic it assumed motivated the purchase. The Xbox line came from a pattern of digital credit purchases. The playlist named “I farted on a rock” came from a household with children in it.
In other words the strange entries are not errors in the sense of bad data. They are the model doing exactly what it was built to do, stated out loud, where the inference chain is visible and slightly embarrassing.
What you can and cannot do about it
| Action | Possible? | Detail |
|---|---|---|
| Read every conclusion Amazon has drawn | Yes | Grouped into sections such as Home & Family and product preferences |
| Delete an individual entry | Yes | Removes it from personalization. Do this for anything inaccurate or too personal |
| Correct an entry | Yes | Feedback on specific details is the intended loop. Fixing beats deleting if the category is useful to you |
| Add things on purpose | Yes | You can volunteer details you want factored in, which is the point of a profile you control |
| Turn the whole thing off | No | There is no single opt-out from personalization itself. Delete the lines, and the model keeps generating new ones from new activity |
| Stop the inputs | Partly | Browsing history can be managed and recommendation preferences adjusted, but purchases and searches are inherent to using the service |
That fifth row is the honest limitation. You can edit the output. You cannot decline the process. It is the same shape as a lot of modern privacy controls, where the visible surface is adjustable and the engine underneath is not, which is exactly the pattern we ran into with ChatGPT’s human review programme, where the off switch existed but sat two menus deep.
The entries people are actually finding
Worth a quick tour, because the range tells you something about how the model generalises.
| What Amazon wrote | The likely chain of reasoning |
|---|---|
| “Has flat buttocks” | Bought butt-scrunch leggings. The model inferred the motivation rather than the purchase |
| “Has never known anyone who has lost a leg in an accident” | Filed under Home & Family. A negative inference nobody asked for and no product needs |
| Too short to reach into a 40-inch box | Almost certainly a product review or a return reason, generalised into a physical fact |
| “Steadily funding Xbox ecosystem with digital credits” | Repeat gift card purchases. Accurate, and phrased like an intervention |
| A child-named playlist in a parent’s profile | Shared household activity leaking across the family account |
The last row is the one with a practical lesson. On a shared Amazon account, the profile is a blend of everyone using it, and some of what is written there will be about somebody else. If your household shares one login, the page is worth reading together rather than alone.
Before you screenshot yours. The entries that have gone viral are jokes about leggings and gift cards. The ones people are not posting are the ones inferring health conditions, pregnancy, religion, grief or money trouble from a purchase pattern, and those are in some people’s profiles too. Read the page before you share the page.
Why Amazon built a page it knew would be mocked
Two reasons, and only one of them is about you.
The first is regulatory weather. Showing people the profile and letting them amend it is the direction privacy law has been pushing for years, and a visible, editable profile is a much better answer to a regulator than an opaque model. The second is data quality. A recommendation engine that lets millions of shoppers correct its wrong assumptions gets cleaner inputs for free, and Rufus, Amazon’s shopping assistant, now personalizes what it retrieves based on this profile. Every deletion is a labelled training signal.
That is not cynicism, it is just the deal. You get to see and fix the file. Amazon gets a better file. Both of those things are true, and the trade is more favourable to shoppers than the version where the profile stays hidden.
It also sits inside a bigger shift in how retail software works, where the interface adapts to context rather than presenting everyone the same shelf. Our explainer on what contextual commerce is and how it changes a storefront covers why personalization stopped being a sidebar of suggestions and became the structure of the whole shop.
A five-minute cleanup, in order
- Open the page. App: Me, your name, About You. Desktop: hover your name above Account & Lists, then About You
- Read all of it, not just the funny section. Work through each category rather than stopping at the first entry that makes you laugh
- Delete anything about your body, health, family circumstances or finances. None of it improves a product recommendation enough to be worth having written down
- Correct the things that are merely wrong. If you are being shown the wrong size, diet or category, fixing the entry is faster than fighting the recommendations
- Add anything genuinely useful. Allergies, sizes and a pet’s species are the sort of detail that makes the output better and costs you nothing to state plainly
- Check it again in a month. New activity generates new conclusions. This is a page to revisit, not a one-off cleanup
The wider point about Amazon and your account
This is the second time in as many weeks that an Amazon account setting has turned out to matter more than people assumed. The company has been working through a round of Prime refunds tied to the FTC case, with a $200 cap and groups of members paid in waves, and plenty of eligible people only discovered it through coverage rather than from Amazon.
The common thread is that the controls and the money both exist, and both are findable, and neither is advertised. Which is a reasonable argument for spending ten minutes inside your account settings once a quarter, whatever the viral screenshot of the week happens to be.
What to watch next
- Whether Amazon adds a master off switch. Public mockery is a strong motivator, and “delete each line individually” will not satisfy regulators indefinitely
- Whether sensitive categories get fenced off. A model that writes about a shopper’s body should probably not be allowed to write about their health
- Whether the page reaches more countries. Availability has expanded in stages since May and is not uniform
- Whether rivals copy it. An editable AI profile is a strong transparency claim, and the first retailer to match it gets the same headline without the flat-buttocks paragraph
- Household separation. Shared accounts blending profiles is a design problem Amazon has not solved, and it is where the genuinely awkward entries come from
The bottom line
Amazon’s About You page is a list of conclusions a model has drawn about your life from your receipts, and it has been quietly accumulating since May. The internet found it this week because some of those conclusions are absurd, which is the best possible reason for a privacy page to go viral.
Go and read yours. Delete the entries about your body and your household. Correct the ones that are simply wrong, because those are the ones making your recommendations worse. And accept the part you cannot change: the page is editable, the inference is not, and every purchase you make from here writes another line.
Sources and further reading
- Amazon: How to view and edit your shopping preferences with About You
- TechCrunch: How to find out what Amazon thinks it knows about you
- Fast Company on Amazon’s shockingly specific customer profiles and how to find yours
- UNILAD Tech: Amazon’s About You AI reveals your shopping secrets
- ZonGuru on the About You profile and what it means for sellers
- 604 Now on the About You notes trending online
About this article: GeekBlog covers U.S. technology news, AI, phones, smartwatches and gaming. Every story is written and checked under our Editorial Policy. Spotted a mistake or have a story tip? Contact our editors.
