TL;DR:
- Resale platforms and marketplaces employ machine learning and AI to combat counterfeits.
- Amazon disposed of 6 million counterfeit goods, while StockX stopped $30 million of fake sneakers.
- Machine learning detects inconsistencies and trains authentication algorithms.
- The Real Real utilizes machine learning for image retouching and product copy.
- StockX relies on 20-50 touchpoints and embedded technologies for authentication.
- eBay acquired 3PM Shield for additional counterfeit prevention technologies.
- Machine learning presents challenges and risks, requiring expertise and quality data.
- Digital IDs and collaborative efforts are crucial in countering counterfeiting.
Main AI News:
Resale platforms and online marketplaces are taking decisive action against counterfeit goods, thanks to the integration of machine learning and artificial intelligence (AI) technologies. As reported in Amazon’s third annual Brand Protection Report, the e-commerce giant eliminated a staggering 6 million counterfeit products in the previous year. Similarly, StockX, a prominent sneaker marketplace, successfully prevented the circulation of nearly $30 million worth of fake sneakers on its platform. The Real Real, specializing in luxury consignment, also played a pivotal role by thwarting 200,000 counterfeits from infiltrating the market since its inception in 2011.
The presence of counterfeits predates the digital era. However, for resale platforms that handle a substantial influx of items daily, detecting every fake item can be an arduous task. Although trained personnel oversee authentication and verification processes, machine learning possesses the ability to uncover inconsistencies that often elude the human eye. Resale platforms frequently leverage machine learning algorithms to train their systems to recognize patterns and distinguish genuine products from counterfeit ones. This involves scrutinizing various aspects such as logo placement, trademarks, and other distinguishing features, allowing these platforms to promptly identify and flag potential counterfeit items.
Since 2018, The Real Real has utilized machine learning to enhance its image retouching and product copywriting processes. In recent years, the company has made significant strides in developing additional tools specifically designed to combat counterfeits. One of these innovations is a system called “Shield,” which collects information about consignors and alerts the platform if a consignor has previously submitted counterfeit items. However, “Shield” does not exclusively focus on the individual product itself. To address this, The Real Real introduced “Vision” about a year and a half ago, utilizing camera technology to aid in the verification process. Currently, The Real Real is in the process of integrating these two systems into a unified solution.
Christopher Brossman, Vice President of Machine Learning at The Real Real, stated that machine learning ultimately provides decision support. While human personnel are involved in performing various tasks to authenticate products, this technology enables them to complete their tasks more efficiently. Any products failing to meet The Real Real’s standards are promptly returned to the sender, in accordance with a three-strike policy. In cases where there is evidence of deliberate intent to defraud, the item is sequestered and handed over to law enforcement.
StockX, on the other hand, exclusively accepts brand-new products and subjects each item to rigorous evaluation based on 20 to 50 touchpoints. These checkpoints include meticulous inspections of packaging, size, color, and embedded technologies such as chips and IDs that manufacturers incorporate into their handbags and shoes. According to Paul Foley, Head of Brand Protection at StockX, counterfeit items emit certain irregularities that allow skilled authenticators to discern their authenticity. Foley also revealed that a significant number of fakes intercepted by StockX are inadvertently listed by sellers who are unaware of their counterfeit nature. In such cases, StockX promptly returns the items to the original senders.
To fortify its authentication processes, StockX has recently established three new authentication centers in Mexico City, Tokyo, and Berlin. Additionally, the company is in the process of constructing an Innovation Lab in Detroit, dedicated to exploring cutting-edge technologies that can further support its authenticators.
EBay, recognizing the significance of countering counterfeit products, acquired 3PM Shield, a solutions provider, in February. This strategic move grants eBay access to additional technologies aimed at preventing the sale of counterfeit goods. A spokesperson for eBay emphasized the platform’s comprehensive approach to protecting buyers and sellers, combining people, policies, and technology. Leveraging artificial intelligence-powered tools, eBay proactively blocked a staggering 88 million suspected counterfeit listings in 2022, and following a meticulous review by eBay investigators, over 1.3 million items were promptly removed.
The battle against counterfeits has intensified for resale marketplaces, particularly as consumers exhibit a growing inclination toward purchasing secondhand goods. This trend is especially prominent in the luxury market, as evidenced by The Real Real’s impressive 22% year-over-year increase in active buyers during its first fiscal quarter. While established marketplaces like eBay have experienced a decline in active users over the past year, new competitors are emerging, enticing shoppers to explore alternative platforms. To stand out and secure sales, many platforms are now placing their bets on machine learning, aiming to prove that their products are unrivaled in terms of authenticity.
Despite its numerous advantages, machine learning presents its own set of challenges. Acquiring and building machine learning systems can be a costly endeavor, and the efficacy of these systems heavily relies on the quality of the data they are exposed to. Platforms starting from scratch must gather a diverse array of authentic and counterfeit products to effectively train their algorithms. Christopher Brossman from The Real Real emphasized the necessity for expertise in programming these tools, particularly in the domain of counterfeit detection. Achieving proficiency in this area demands a deep understanding of the subject matter.
Furthermore, like any technology, machine learning is not infallible. Kassi Socha, Director Analyst at Gartner, highlights that the output of AI-enabled machine learning is not always predictable, as models continue to learn and evolve. Consequently, exposure to incorrect or insufficient data can compromise the system’s ability to detect counterfeits, posing significant risks and challenges for these platforms. Socha also warns of the possibility that counterfeiters could exploit machine learning to refine their illicit operations, underscoring the urgent need for industries like apparel, footwear, and accessories to invest in digital IDs capable of thwarting such attempts. Certain companies, such as Mulberry and Prada, are already taking proactive measures by introducing digital IDs, NFC, and RFID tools into their products. Collaborative efforts are also being undertaken, such as the Aura Blockchain Consortium initiated by luxury retailers like LVMH and Cartier, which offers a certificate guaranteeing the authenticity of luxury items.
Paul Foley from StockX calls for increased collaboration among platforms and brands to collectively combat counterfeiters. By pooling their extensive datasets and knowledge, these stakeholders can synergize their efforts and bring about a game-changing impact in the fight against counterfeiting.
Conclusion:
The integration of machine learning and AI technologies in resale platforms marks a significant step in the battle against counterfeits. Platforms like Amazon, StockX, and The Real Real are successfully using machine learning to identify inconsistencies and train authentication algorithms. This ensures a higher level of accuracy in distinguishing genuine products from counterfeit ones. However, challenges exist, such as the need for extensive and high-quality data, expertise in programming, and the potential for counterfeiters to exploit machine learning. The market’s response to counterfeiting through machine learning and the adoption of digital IDs and collaborative initiatives demonstrates a commitment to safeguarding consumer trust and reinforcing the authenticity of products in the resale industry.
