Company Develops AI Tech That Spots a Fake Designer Product From Single Photo
A company has developed an AI image-recognition model that can identify a fake designer product from a single photo — with luxury fashion brand Lacoste becoming its first customer.
According to a report by Semafor, a French start-up called Vrai AI developed the technology after training its image-recognition model on thousands of photos of genuine designer goods.
The company claims the technology can combat the counterfeit black market, which makes up an estimated 2.5% of global commerce and costs businesses billions annually.
Semafor reports that Vrai AI launched its software earlier this month and its first client is Lacoste, the designer brand known for its iconic crocodile logo, which often makes it a prime target for counterfeiters.
David G. Stork, the company’s chief scientist and a Stanford University adjunct professor, tells the news outlet that a counterfeit Lacoste product having obvious flaws like extra teeth or an overly long tail in its signature crocodile are obvious giveaways that an item is fake.
Stork says that Vrai AI takes a more nuanced approach, analyzing subtle details that are harder to detect. Stork, who is renowned for his work at the intersection of AI and art, has applied similar techniques to uncover forgeries and identify uncredited contributions in historical paintings.
Stork explains that Vrai AI’s model can differentiate between natural manufacturing variations—since even machines don’t produce identical logos every time—and counterfeits. Subtle details, such as a slightly incorrect shade of green, an over-rotated crocodile eye, or stitching spaced too far apart, can trigger the AI’s sensors.
Vrai AI claims its technology is 99.7% accurate with Lacoste items.
Lacoste has adopted Vrai AI to combat return fraud, where scammers buy authentic products online and return counterfeits, which can sometimes end up being resold to unsuspecting customers.
With Vrai AI’s app, warehouse staff can photograph returned items to instantly verify their authenticity, helping Lacoste save money and protect its customers from counterfeit goods.
According to Semafor, Vrai AI also plans to extend its technology to tackle counterfeit medicines — a crisis responsible for hundreds of thousands of deaths annually. Its AI model could detect subtle visual discrepancies, such as variations in font size or color on medical packaging.
To train its algorithm, Vrai AI would need around 50,000 images, which could come from a collaboration with a major pharmaceutical company or from the company capturing its own photos.
Image credits: Header photo licensed via Depositphotos.
