The Netherlands has one of Europe’s most active second-hand marketplaces. Marktplaats attracts millions of visitors every month. Facebook Marketplace has become the first stop for buyers hunting a used bicycle, laptop or car. And in that traffic, fraudsters are waiting.
The scheme is always the same: a tempting price, convincing photos and a seller who asks for a deposit before you can pick up the item. What most buyers do not know is that the photos are stolen. And that is precisely the vulnerability that photo geolocation exposes.
How second-hand marketplace fraud works
A fraudster placing a fake listing takes almost no risk if the photos are chosen carefully. They copy images from a real listing, ideally from a platform in another region or a foreign site, and attach them to a new listing with a different location.
The buyer sees a clean photo of a road bike in Amsterdam, gets in touch and is told that collection is unfortunately not possible but the bike will be shipped after a deposit of 150 euros. The 150 euros disappears, the bike does not exist and the photo was actually taken in Antwerp or Berlin.
This type of fraud is widespread. The Fraudehelpdesk registers tens of thousands of reports of internet fraud annually, and marketplace fraud makes up a substantial share. Individual losses range from tens of euros for a phone to several thousand euros for a car or motorbike.
The photos reveal the location
Second-hand listings almost always include exterior photos. A bicycle is parked on the street. A car sits on a driveway or in a car park. A campervan stands next to a house. Those backgrounds contain location signals: facades, paving, street furniture, vegetation and surrounding buildings.
GeoPin analyses these backgrounds and estimates the most likely location in the Netherlands based on visual characteristics. The system is trained on street-level imagery and reference data specific to the Dutch context, from the distinctive terraced housing of Randstad neighbourhoods to the building density of provincial towns.
If a listing claims the item is in Utrecht but the background matches a residential area in Groningen, the discrepancy becomes visible within seconds.
A practical scenario
Suppose a listing on Marktplaats offers a light-coloured motocross bike for 2,800 euros, with the location given as Eindhoven. The photo shows the bike parked beside a brick facade in a recognisable style.
A buyer who uploads the front photo to GeoPin receives a predicted location that places the property in a residential street in Venlo, 75 kilometres from the stated address, with a confidence score of 0.82.
Three possible explanations: the seller has a collection address different from the listing address, the bike is temporarily stored elsewhere, or the photo was taken from another listing. In any case, further investigation is warranted. For the buyer, this is the moment to stop.
How platforms can use this
Marktplaats and similar platforms process hundreds of thousands of new listings every day. Manual verification is impossible. But automated photo geolocation as part of the upload workflow is scalable.
Integration via the GeoPin API works as follows:
Step 1: Address extraction. The platform retrieves the stated location from the listing and geocodes it to coordinates.
Step 2: Photo geolocation. The uploaded photos are sent to the GeoPin API. The system returns an estimated location and a confidence score.
Step 3: Distance check. If the distance between the predicted photo location and the stated address exceeds a configured threshold, such as 50 kilometres for items typically sold locally, the listing is flagged.
Step 4: Action. Depending on severity, the platform can hold the listing for review, ask the seller for additional verification, or at high confidence scores block the listing automatically.
The cost per API call is minimal compared to the damage caused by even a handful of fraudulent listings per day.
What buyers can do right now
Not every platform integrates verification technology. Individual buyers can protect themselves by following a few simple steps.
Reverse image search. Run the photo through Google Images or TinEye. If the same image appears elsewhere, ideally with a different location or seller name, something is wrong.
GeoPin for individual use. The API is available directly at geopin.nl for individual queries. Upload the exterior photo of the listed item and compare the returned location against the listing address.
Insist on collection. If a seller refuses to let you view the item in person before purchase or makes collection impossible, that is a strong warning sign. For high-value items, inspecting in person is the only real certainty.
Use protected payment methods. Credit cards and platform-backed payment systems offer buyer protection. A direct bank transfer to an unknown party is almost impossible to reverse.
The broader picture: AI as platform responsibility
The Digital Services Act obliges large platforms to have systems for reporting and removing illegal content. But prevention, stopping fraudulent listings from going live in the first place, falls outside the strict obligations. Yet prevention is economically and reputationally more attractive than reactive removal.
Photo geolocation is a concrete preventive measure that fits into the validation workflow of any second-hand platform. The technical barrier is low, the costs are manageable and the effect on fraudsters is direct: their method, stealing photos from listings in other locations, stops working.
For buyers, awareness is the first line of defence. The photos do not lie, but they lie about where they come from. Photo geolocation makes that difference visible.
GeoPin is an AI photo geolocation service optimised for the Netherlands. Read more about how GeoPin works or explore the API integration guide.