Vinted, Leboncoin: how AI photo scams are exploding

As a Vinted seller for several years, I've watched the fake-return scam evolve radically with the arrival of generative AI. What once required photo-editing skills is now accessible to anyone in a few seconds. The result: an explosion of photo fraud on C2C marketplaces, and sellers who lose money with no recourse.

How does the fake-return scam on Vinted work?

The most common scenario has become a classic that every regular seller knows. You sell an item in good condition, a piece of clothing, an electronic device, a collectible. The buyer receives it, sees that it matches the listing, but decides to obtain a fraudulent refund. They photograph the item in good condition, then use an AI tool to generate an altered version of the photo: a stain added to a garment, a scratch on a screen, a non-existent defect on an object. They then open a dispute with the marketplace claiming the item arrived damaged, with a "proof" photo. The seller is trapped: the platform sees two contradictory photos and, when in doubt, usually rules in favor of the buyer. The seller loses both the item and the money.

What are the new AI photo-fraud techniques?

Fraudsters now use several techniques made trivial by generative AI. Inpainting lets you modify a specific area of a photo by describing the desired change in text: "add a tear to the shirt collar" is enough to generate an ultra-realistic defect. Outpainting extends an existing photo to show a different context, for example an opened parcel with visible damage that doesn't exist. Full generation makes it possible to create from scratch a photo of a product in degraded condition, using the original listing photo as a reference. Face swap in ID photos facilitates identity theft for sales of regulated products. All these techniques are available for free or for a few euros a month via consumer mobile apps.

Why can't the platforms block this fraud?

Vinted, Leboncoin, eBay and other marketplaces face a structural problem. Their dispute-resolution system relies on comparing photos supplied by both parties. But an AI-altered photo is often indistinguishable from an authentic one, even for an experienced human moderator. Platforms also can't require every user to prove the authenticity of their photos, as this would add considerable friction to the selling process. Some platforms have started experimenting with AI-image detectors, but as we've seen, these tools have significant false-negative rates on the latest generation models. The volume of transactions also makes manual verification impossible: Vinted processes millions of transactions a month in France alone.

How much do sellers lose because of these scams?

Precise figures are hard to come by because many sellers don't report losses they consider too small to justify action. However, on seller forums and Vinted groups, testimonials keep multiplying. Sellers report losses ranging from 20 to 500 euros per scam, with some recurring cases where the same seller is targeted several times. At the platform level, Vinted acknowledged in 2025 that photo-related disputes accounted for a growing share of its mediation costs, without giving precise numbers. For a casual seller, a single scam can be enough to discourage any future activity on the platform. For professional sellers making dozens of sales a month, cumulative losses can reach several thousand euros a year.

How can you protect yourself in practice as a seller?

The first line of defense remains systematic documentation. Photograph each item from every angle before shipping, with a close-up of areas likely to be disputed (seams, screen, corners). Include a dated element in the photo, that day's newspaper, your phone's screen showing the date, even if it can be faked. Film the packaging of the parcel through to sealing and handover to the carrier. However, all this documentation remains fragile evidence against a fraudster who uses AI to generate counter-evidence. The only structurally solid solution is cryptographic certification at the moment of capture. CertiPix does exactly that: every photo taken with the app receives a timestamped, tamper-proof certificate that proves when the photo was taken and that it hasn't been altered. In a dispute, the certificate is irrefutable proof that the product was indeed in the documented condition at the time of shipping.

The question is no longer whether you'll be the victim of a marketplace photo scam, but when. Certification at the moment of capture is the only structurally reliable protection.