Fake Photos on Marketplaces
Generating a photo of an item that does not exist now costs nothing and takes seconds. Meanwhile the oldest listing scam — stolen photos of someone else's real item — still works fine. Two different problems, two different checks, both quick.
The three kinds of fake listing photo
Generated. The item was never photographed because it does not exist. Detected by generation analysis: flat noise, even detail, no camera data.
Stolen. A real photo of a real item, belonging to someone else. Forensics finds nothing wrong, because nothing is. Reverse image search catches it instantly — it will usually appear on a manufacturer page or another listing.
Edited. A genuine photo of the actual item with the damage removed. Shows as a region with its own compression and noise history, which is what manipulation analysis looks for.
Signals specific to product photos
Text that isn't text. Brand names, model numbers, spec panels and barcodes degrade into letter-shaped forms. On packaging this is usually obvious once you zoom.
Impossible construction. Seams that do not meet, hinges that could not move, cables entering nowhere, screws in decorative positions.
Studio quality from a casual seller. Catalogue lighting on a supposedly hand-taken photo, with no camera metadata behind it.
Materials that drift. A weave, wood grain or knurl that changes pattern across one surface. Physical materials do not do that.
One angle only. Generated listings rarely have consistent multiple views of the same object, because consistency across angles is hard.
What to ask the seller
This is more effective than any tool, and it takes one message.
Ask for a photo with something specific in frame — today's date written on paper next to the item, or a particular angle you name. A real seller sends it in a minute. A fake one stalls, sends the same picture again, or disappears.
Ask for the original file, not a screenshot. Marketplace uploads are re-compressed; the original preserves metadata and gives a far better check.
Ask for a short video turning the item over. Still much harder to fake convincingly than a still image.
Refusal on any of these tells you more than a detector score would.
Running the check
Reverse image search first — it catches stolen photos in seconds and is the most common case by far.
Then the product photo check for generation and editing. Expect lower confidence on images pulled from a marketplace, since the platform re-compressed them on upload.
If you are buying something expensive, do both and ask for the extra photo. The combination costs five minutes and covers essentially everything except an outright lie about condition.
Key points
- Three cases: generated, stolen and edited — each needs a different check.
- Reverse search first: stolen photos are the most common listing fake.
- Watch for broken text on packaging and materials that change pattern.
- Ask for a photo with today's date in frame — refusal is the answer.
- Marketplace compression weakens analysis; request the original file.
Frequently asked questions
Check an image yourself
Reading about it only goes so far. Our AI image detector runs twelve independent methods and shows you every score, not just a verdict.
More from the blog
Try it on your own image
Free account, 3 checks included, no card required.