A quick, honest caveat to start with: an AI-generated product image is not automatically a scam. Plenty of legitimate brands now use CGI and generative tools to stage a lamp in a room they never shot, or to render a color variant they haven't manufactured yet. So the question isn't really "was this made with AI" β it's "can I trust that a real product matching this photo will actually arrive." The checks below answer that second question, which is the one that costs you money.
Bellingcat's investigators and consumer reporters at outlets like Tom's Guide have documented the same shift: fake storefronts increasingly skip the work of stealing photos and just generate a plausible-looking product on demand. The tell is rarely one smoking gun. It's a synthetic-looking image sitting next to a brand-new seller, no buyer photos, and a price that's too good β read the whole picture, not one pixel of it.
The five-step check
Zoom in first β the tells hide at full size
Open the largest version of the image the listing offers and look closely. The single most reliable giveaway is text: AI still struggles to spell, so packaging, labels, buttons, and stitched-in brand names come out as confident gibberish that looks like writing until you actually read it. After text, check any hands holding or using the product β fused fingers and extra knuckles remain common. Then trace the product's own edges and hardware: zippers that go nowhere, seams that dissolve into the background, screw holes that don't line up, a strap that merges into a shadow. Real product photography is boringly consistent; AI renders drift.
Reverse image search the main photo
This is the step that settles most cases, and the FTC recommends it outright for marketplace buying. Drop the listing's hero image into Google Lens and TinEye. Three outcomes matter: if the same photo appears on dozens of unrelated listings, it's a reused template, not the seller's actual stock; if it traces back to the real brand's site or a stock library, the seller lifted it; and if it returns nothing at all, that's consistent with a freshly generated AI image made for this listing. TinEye's oldest-first sort helps you find where a photo really started. Archived copies of older listings can also surface the negative reviews a scam store has since deleted.
Compare the hero shots against real buyer photos
Filter the reviews for customer-uploaded images. Photos shot by actual buyers look like it β kitchen counters, uneven lighting, a hand in frame, the box on a real floor. When those match the listing, confidence rises fast. When a listing has nothing but flawless studio renders, no buyer photos at all, and a wall of five-star text reviews, treat the product as unverified rather than fake β but do not send money on the strength of the hero image. The FTC's own advice is blunt: if you can't see photos of the actual item, you can't judge its condition or even be sure the seller has it.
Read the seller and the price, not just the picture
AI product photos rarely travel alone. They tend to sit on brand-new stores with no history, listings with no seller contact information, specs that stay vague on purpose, and a price that undercuts everyone else by an amount that doesn't make sense. Any one of these is weak on its own; stacked together with a synthetic-looking image, they're the pattern consumer-protection agencies describe. Prefer sellers who accept payment methods with buyer protection, like credit cards, so a bad order is recoverable.
Check the product against the real brand
If a listing claims a known brand, go to that brand's official site and find its list of authorized retailers. Compare the official product photos against the listing's β a genuine product page will have consistent, real photography and matching model numbers. A brand you can't find a website for, or a 'brand' that exists only inside marketplace listings, is its own answer. This is also where you separate a bargain from a counterfeit: the same image checks that catch AI fakes catch stolen real photos wrapped around a knock-off.
Tools worth having open
None of these decides it for you. Reverse search plus a look at real buyer photos beats any single detector score β run two or three and weigh what agrees.
| Tool | Type | What it's good for |
|---|---|---|
| Google Lens lens.google.com | Free | Fastest reverse search from a phone or browser; finds the same photo on other listings and stock sites. |
| TinEye tineye.com | Free web tool | Exact-match reverse search with an 'oldest' sort β good for finding where a product photo first appeared. |
| Sightengine / WasItAI sightengine.com Β· wasitai.com | Free web demo | AI-image probability as a second opinion β a flag to investigate, never a verdict on its own. |
| Marketplace review filter the listing's reviews tab | Built into the platform | Surfaces buyer-uploaded photos so you can compare real-world shots against the hero renders. |
| Official brand site the manufacturer's own domain | Free | Confirms real product images, model numbers, and which sellers are actually authorized. |
Where these checks mislead you
The trap is treating "looks AI" as "is a scam." A few limits to keep you honest:
- Big retailers use CGI too. A rendered lifestyle shot on a major brand's own page is normal marketing, not fraud. Provenance and seller trust decide it β not whether a computer touched the image.
- Fresh AI images have no reverse-search history. An image generated for one listing may return zero matches, which looks clean but proves nothing. Absence of copies is a reason to lean harder on seller and buyer-photo checks, not to relax.
- AI detectors false-positive on real product shots. Heavy retouching, white-background cut-outs, and platform re-compression can all read as "likely AI." Treat a score as a flag to keep looking, never as the answer.
- Stolen real photos are the older, still-common trick. The same reverse search that catches AI catches a genuine photo lifted from the real brand and wrapped around a counterfeit. Either way, the fix is the same: verify the seller.
The habit that saves the money
When a deal makes you want to buy right now, that urgency is the thing to distrust. Take the ten seconds to reverse image search the photo and scroll to the buyer pictures before you check out. A real product will still be there tomorrow at roughly the same price; a listing built on a generated image and a countdown timer usually won't survive that pause. Slow down at the exact moment the page is trying to speed you up.
Checking a claim, not just a photo?
FAXTR searches 100+ fact-checking organizations in one query β free, no login required.