AI counterfeit detection rests on a simple but powerful idea: a machine doesn’t “feel” that a bag is genuine the way an expert does. Instead, it compares thousands of tiny details against millions of reference images in seconds. At the heart of this shift is a US company called Entrupy, whose annual “State of the Fake” report has become a reference point for anyone working in luxury fashion and resale.

In this article we look at the key findings of the 2026 report, explain how AI “sees” what the human eye misses, and, most importantly, what it means for you as a designer building your own label.

What is Entrupy and how does it work?

Entrupy is a New York-based company that built a system for examining products through magnified images of their finest details: the texture and pores of leather, the shape and regularity of stitching, the sharpness of print, and the way a logo is stamped on hardware and lining. Its AI then compares those details against a vast library of images of genuine and counterfeit items and returns a result within a very short time.

The system originally needed a small microscope device attached to a phone. But the company announced that from 18 May 2026, bags and leather goods can be authenticated directly from a smartphone, with no extra hardware. Authentication is no longer reserved for experts and large resellers.

What this means for you: every small detail of a garment, from stitch density to how accurately a print matches at the seams, has become a “fingerprint” that can be read by a machine. Quality is no longer just a matter of taste; it is proof of identity.

What did the “State of the Fake 2026” report reveal?

The report was released in April 2026 and covers data from 2025. Over that year, Entrupy processed products worth more than $3.7 billion. The vast majority came back authentic, while around 8% could not be authenticated.

The most targeted brands were no surprise: Louis Vuitton, Gucci, Prada, Chanel and Dior. More striking is the report’s observation that the fake economy is fluid: when one brand becomes harder to copy, counterfeiters move on to an easier target. Loewe, for example, topped last year’s risk list but disappeared from it this year as awareness and authentication volumes grew.

Chanel offers an important lesson: despite being one of the most sought-after brands, its fake rate stayed relatively low, which the report attributes to its exclusivity.

Apparel and footwear: where fakes hide

Handbags grab the headlines, but the report shows that sneakers are more likely to be fake than bags. The bigger surprise came from clothing: when Entrupy expanded into apparel, roughly one in three items could not be authenticated.

Fear of God Essentials topped the list of riskiest apparel brands, with the overwhelming majority of submitted pieces unverified. The reason, according to the report: accessible, high-demand basics with simple branding are the easiest targets for counterfeiters.

What this means for you: a design whose value rests on its logo alone is the easiest to copy. A design that carries its value in the cut, the detailing and the quality of finish is much harder for a counterfeiter, and much easier for any smart system to recognise as genuine.

How does AI “see” what the eye misses?

A skilled counterfeiter can copy the overall look, but rarely reproduces everything precisely. AI looks specifically for those small gaps: stitches per centimetre, the direction of the leather grain, the thickness of printed ink, and the spacing between letters in a logo.

Interestingly, these are exactly the details we learn in pattern cutting and sewing: matching a print across seam lines, keeping stitch length consistent, and finishing the inside cleanly. What used to be called “craftsmanship” is now measurable data.

What can future designers take from this?

If you are building your own label, here are three practical steps inspired by the report:

1. Document your pieces from day one. Photograph the stitching, labels and internal finishes of every approved sample. This archive becomes your reference if copies ever appear.

2. Make a detail your signature. A distinctive construction detail, such as a particular edge finish or lining arrangement, is harder to copy than any logo.

3. Protect your name legally. Registering your trademark and protecting your designs should happen before your product succeeds, not after.

Conclusion

AI counterfeit detection doesn’t only protect the big houses; it is changing the very meaning of quality. In a world where every stitch can be read by a machine, precise craftsmanship becomes the strongest protection a designer has. The question worth asking: do your pieces carry details that make them impossible to copy?

Source: Entrupy’s annual “State of the Fake 2026” report.