Man used AI to fake spoiled food in photos and obtain refunds from stores

In China, a man turned a neural network into a tool for defrauding online retailers. For months, he bought expensive fruit on marketplaces and then used AI to edit photos so that they appeared to show rotten and spoiled areas. He then demanded refunds while keeping the fruit for himself.

According to Chinese IT publication IT Home, the scheme primarily targeted expensive products — in particular, durians and cherries. After receiving an order, the man photographed the product and then used a neural network to add signs of spoilage to the images. He used the fake photos as evidence that the seller had allegedly sent him an inferior product.

After the refund was processed, the money was returned to the buyer, while he did not throw away the fruit. Instead, he resold it online, effectively making a double profit: the funds refunded by the marketplace and the money from selling the goods he had received.

Over four months, he placed more than 100 orders worth a total of around 16,000 yuan. However, the scale of the unusual «business» attracted the attention of the marketplaces. Due to the large number of complaints related to similar refunds, the entire county where the suspect lived was even blacklisted.

Investigators tracked the man down through the recurring delivery addresses. During a search, police found 75 mobile phones in his possession, which investigators believe were used in the fraud scheme.

As a result, the court found the man guilty. He was sentenced to one year in prison and fined 5,000 yuan for manipulating the refund system.

The case highlighted a new problem for marketplaces: neural networks make it possible to create convincing images of spoiled goods within minutes, meaning that the familiar refund system based on photos submitted by buyers is becoming far more vulnerable.