Can AI end the era of the fashion super-fake?

“The bag she wanted to sell with us was an Hermès Mini Kelly – it was visually perfect,” says Hanushka Toni, CEO and co-founder of luxury resale company Sellier: she is speaking of a potential vendor. “However, when we inspected it, we noticed that it didn’t smell like an Hermès bag – which has a distinctive ‘new-car smell’…” Sellier ran the Mini Kelly through its new AI software. “It was confirmed as a fake.”

Fashion is adapting to the rise of AI in multiple ways. Resale company Hardly Ever Worn It, HEWI, recently launched the search-optimisation chatbot “Maia” on their platform, with the goal of saving customers hours of scrolling for a certain look or product; Shopify has also recently integrated ChatGPT into its Shop app, assisting sellers with product descriptions. But the fight against “super-fakes”, which have long tormented luxury brands and resale companies, is the main priority.

The global luxury counterfeit and replica market is worth an estimated $1.9tn, with black-market manufacturers using increasingly advanced techniques such as laser cutting and 3D printing to confound even the creators of the original product. With AI technology, luxury brands and resale companies have armed themselves with the data-fuelled authentication systems they believe will finally outwit them. These systems essentially compile a digital encyclopaedia of counterfeit and authentic article properties, to then spot tiny inconsistencies in fabric, stitching or metals. LVMH-owned brand Patou recently launched AI-verification system Authentique Verify, effectively digitally “chipping” their items at the very beginning of the production chain, allowing them to track them and prevent any return fraud (where fake items are sent back in lieu of the original). Photographs of the authentic products, taken on a phone, process all their microscopic properties – which, like fingerprints, allegedly cannot be replicated. 

We simply can’t risk shipping out products that are not authentic

The scale at which most conglomerates work makes experimenting with brand new counterfeit technology risky. For smaller luxury brands and resale businesses though, trialling the technology is a no brainer. Sellier, a global resale platform founded in 2019, specialises in Hermès, Chanel and Louis Vuitton bags and accessories, with spends being upwards of £60,000, and a clientele based in Knightsbridge and Monaco. Toni felt it was essential to invest in AI when one client was stung by a fake handbag she’d sourced from an unnamed competitor. “It’s the most gut-wrenching thing to have to tell people, as these are not small purchases and it’s often a challenge to get your money back.” 

The company uses expert human authentication, but it’s not enough now, she continues. “With a bad fake, human authentication will get you there, but with a super-fake, there is still a 15 per cent chance for error.” What’s more, for a small company like Sellier, each transaction has to be water-tight. “With our low-volume, high-margin strategy, we simply can’t risk shipping out products that are not authentic.” 

Entrupy works by sending you a smart device (a microscopic attachment) for an iPhone SE, if luxury handbags need authenticating; for trainers/sneakers, it’s simply a case of downloading Entrupy’s app on an iOS device. The app then scans and photographs the item before filing through millions of data points across various images to arrive at a judgement. The software’s edge is its age: it has been ingesting the properties of thousands of both authentic and counterfeit products since 2012 to build an unrivalled database. Entrupy’s CEO and co-founder Vidyuth Srinivasan says that it has “scanned” $2bn worth of luxury handbags to date, with a remarkable 9.6 per cent of those turning out to be counterfeit items.

For brands using fair labour and quality goods this is a plus

Sellier now relies first on an in-house team inspection, then a secondary brand expert, with Entrupy the third and final authenticator. Resale expert and consultant Graham Wetzbarger describes this approach as a “no brainer”. He has authenticated more than $1bn worth of luxury goods himself through his company, Luxury Appraisals. “If you’re doing old Louis Vuitton wallets at £100 then it’s maybe not worth it,” he says. (Sellier’s average basket price is, in contrast, £2,000-£4,000.) He also believes there should be a tailored authentication approach to each company’s business model. Global luxury consignment marketplace The RealReal, for example, has set a goal of 40 per cent AI-powered authentication in 2022, but for handbags only, since that market is so vulnerable to fakers.  

Then there are the recent environmental regulations implemented in France, which subject brands’ supply chains to rigorous scrutiny and are hoped will provide a further blow to the counterfeit industry (particularly since the legislation is predicted to extend beyond France’s borders). “Finally, the consumer will be able to see how the sausage is made,”’ says Wetzbarger. “For brands using fair labour and quality goods this is a plus.”

Pierre Denis, former CEO of Jimmy Choo and an investor in the luxury authentication space, cites blockchain (NFC or NFT code bars sewn into bags) as a preferable method to AI authentication for some brands. It’s a technology that Louis Vuitton, Chloé and Prada are currently experimenting with.

Denis foresees an emerging luxury landscape where brands bring their supply chains in-house, as well as resale, and again apply tailored solutions. But he blends cautious optimism about what AI can do with a warning about the counterfeiters’ relentless innovation. “Everything can be forged,” he explains. “There are hackers in every domain.”