Can AI Be Used to Detect Fake Luxury Goods? Exploring the Role of Technology in Authentication

In the fight against good quality counterfeit luxury goods, the potential harm is largest at the last mile: the collector who believes they are buying a genuine product, but is being deceived. 

As good quality counterfeits become increasingly sophisticated through advanced development and production processes, a question arises: can AI be used to quickly detect these fraudulent products and prevent significant losses, or does the human-centric approach still hold the upper hand?

Let’s take a closer look at the effectiveness of AI in its current state for detecting fake luxury goods and the enduring value of human expertise.

1. Speed: Can AI Authenticate a Product More Quickly?

Traditionally, identifying counterfeits has been a slow-moving process. A collector seeking another opinion needs to either physically take a product to an expert or hire a remote authenticator, who may not immediately be available. This process may take a couple of hours to a few days, depending on the method and availability.

In contrast, AI technology claims to expedite manual processes. Machine learning algorithms can scan millions of web pages, including the brand’s own website, social media, and marketplace listings in real-time, learning an authentic item’s characteristics across a large sample size more quickly than a human could.

These programs are also always online, 24/7, unlike a traditional authentication process. This may make the use of AI interesting for individuals who need to decide in a short period of time.

2. Accuracy: Is AI More Accurate?

While speed may be an advantage, the AI program’s training data may be considered a black box; it is not certain that the program has understood all the edge cases or intangibles that could make a watch listed for sale considered risky. 

A recent example in collector circles was an eBay seller who specialized in watches of a known brand. This storefront continued to sell for some time with positive reviews, until it was found that they were selling counterfeit watches visually very close to the original. 

A closer look at the watches that were listed, and specifically their warranty cards, would have revealed that the certificates were stamped from a shop that had never been an authorized dealer for that brand. Counterfeiters had chosen it at random as it looked convincing, and the sales came in as a result.

This is just one of many examples of an edge case, including short production runs of a watch that may in fact be authentic, but the AI had never seen it before in the training data, or even something as simple as a poor camera angle providing a “fake” result when the watch may in fact be authentic.

While AI can deliver an answer more quickly, speed doesn’t always mean the result is correct. Human experts have intuition and a deep understanding that machines cannot match. They consider the brand’s overall craftsmanship, back catalog, and the finer details of what an authentic watch should have that a machine might miss.

3. Responsibility: Who is Accountable for an Authentication?

While an AI program may offer a scalable authentication solution that could claim to reduce the need for human experts, an element of taking accountability is missing compared to traditional approaches.

Trusted dealers and independent authenticators often have longstanding reputations and credibility within the market. Their standing in the community provides assurance when verifying authenticity, as if there does end up being an issue, they are real individuals and businesses that have given their approval on a product. To put it in other words, there is legal recourse should the product end up being counterfeit.

When using AI, the buyer is relying on a program that is absolved from any responsibility for the result; the collector would have no one to blame but themselves.

The Future of Authenticity: Balancing AI and Human Expertise

While AI offers instant gratification with a result of “good” or “no good,” it is not ready to be relied upon when real money is involved.

As the landscape evolves, AI may improve, and a balanced approach that incorporates machine learning with the wisdom of experienced professionals may have utility for a reliable “first check” in safeguarding authenticity in the luxury goods market. 

For the time being, however, it is not practical to rely on it for authentication, and the value of trusted human experts in this process is still important.

40 Years of Experience in Protecting Intellectual Property. At Selective Trademark Union (STU), we know how important intellectual property (IP) is to brand owners, and we are dedicated to safeguarding it. Check our website at www.stu.net for more information.