We are in a world with rapidly involving tech, where payments are increasingly electronic and fraud attempts are all but inevitable. Fortunately, numerous anti-fraud solutions have appeared in recent years, including Kount’s new AI-driven fraud prevention solution, which uses machine learning to accurately prevent and stop known and emerging fraudulent transactions. According to Kount, its AI is next generation and significantly improves the accuracy of fraud prevention. To learn about this offering, Loyalty360 recently spoke with Gary Sevounts, Chief Marketing Officer for Kount.
He first discussed three trends identified by Kount that led the company to develop its new AI solution. “One trend,” Sevounts said, “is that companies, all the merchants as well as companies with an online, e-commerce presence, are continuously engaging in digital innovation—should that be loyalty programs, e-gift cards, or new types of initiatives to enable a better user experience, enable new revenue channels, or just expand their business. The reason that matters is that, along with new ways of generating revenue, it also introduces new types of vectors for threat, for abuse, for fraud. And fraud is evolving. These companies are trying to protect against and stop fraud.”
He continued, “The second interesting trend is that fraud is becoming more sophisticated. The attackers themselves are beginning to use machine learning, and attacks are not just on payments anymore. They’re also on identities. Things like credential stuffing, bot attacks, phishing, and spear-phishing. There’s a lot of new, automated types of attacks that are targeted, and their use of machine learning makes them more and more effective.”
The third trend Kount has identified is the scope of fraudulent activities. “Not only are attacks more sophisticated now, harder to detect, taking over accounts,” Sevounts said, “but they’re also global. We’re seeing more attacks from around the world.”  
Sevounts argued that these three trends are changing the landscape of threat and fraud prevention very rapidly. Eyeing this shifting landscape, the Kount team saw that putting an AI analogue of an experienced fraud analyst in place could help detect fraud more quickly. This is how the solution began.
According to Sevounts, the product, similar to a flesh-and-blood fraud analyst, looks at historic data. He said, “First, [fraud analysts] look at the historic data. Second, they look at anomalies. Third, given the historic and anomaly views, they weigh those factors and, basically, figure out to what degree they’re comfortable with the transaction.” The solution emulates this process.
He noted that many older products on the market only fulfill one or two of the three steps, failing to take anomalies into account and not always deciding based on degree of comfort. According to Sevounts, other, newer solutions “provide supervised machine learning, where the machine learns from all this information via networks and provides a score.” However, he said, “That’s not artificial intelligence, even if people sometimes call it that. It’s really machine learning.”
Sevounts indicated that the company’s new solution is a step above competing tech. He said, “One of the big differences is that our next gen solution is using AI in a way that enables companies to drive desired business outcomes. It can help companies reduce chargebacks, reduce manual reviews, and improve user experience.”
Sevounts said that the solution will prove useful to a variety of organizations. He noted that most of Kount’s clients are brands, but he also said that Kount has partners that are payment processors and payment providers. Clearly, then, the solution does have multiple uses.
He concluded, “The big value of the AI is not only that it’s significantly more accurate in identifying fraudulent activity versus non-fraudulent activity, but it also emulates an analyst for the merchants. They don’t have to have their fraud analysts constantly reviewing every transaction. Instead, they can rely on it to further fine tune, based on their outcome.”

Recent Content