The fraudsters are getting ever bolder — and attacking businesses and financial institutions (FIs) at scale. Part of their secret to success is their ability to adapt. Which raises the question: How can the good guys adapt?
They can. Andrew Stucchio, global head of network payments pricing, analytics and controls at Discover® Global Network, told PYMNTS, “We recognize that cybercriminals are leveraging technical advances — like bot attacks — in order to test and probe vulnerabilities within payment networks.”
The key to doing battle against the criminals is to adapt, Stucchio noted. The world is evolving, Stucchio said, who added that it’s no longer possible to rely on a single technological tool, or to take fraud-fighting initiatives wholly in-house.
The financial services sector, he said, must leverage some of those same technologies, namely artificial intelligence (AI) and machine learning, to fight fire with fire. Discover Global Network is expanding its platform approach and its processes to help acquirers, issuers and payments networks proactively.
“As a payments network — where we sit in between both issuers and acquirers — we see ourselves as uniquely situated to get a true full view of the global trends and to then harness our data to power AI and machine learning models to fight the fraud on our system,” Stucchio said.
“We’re doing this by leveraging our advanced analytics to detect authorization or settlement data anomalies that could suggest fraudulent behavior,” he continued, adding that “We are combining these very powerful data analyses with other technical advances so that we can more quickly detect the fraud, investigate it, remediate it, and prevent it all for the purposes of creating a better and more insulated network.”
Delving a bit more deeply into some of initiatives of Discover that relate to card-not-present or AI-powered fraud attacks, Stucchio said the company has been creating multiple data-driven rules that analyze transaction-level data throughout the lifecycle of a transaction from authentication, authorization, settlement, the fraud event itself, and then subsequently the disputes.
“The rules that we’ve built are based on historical short- and long-term data across both the merchant and the issuing side of our house, in order for us to detect those unusual transaction patterns that could suggest something fraudulent is happening,” he told PYMNTS.
With those rules in place, Discover has created thresholds to signal when there is an unusual transaction count or a “map” of some unusual combination of data, and the network creates alerts to help flag what might be seen as a potentially risky transaction for merchants and issuers operating on the Discover network.
The process has been an iterative one, where Discover created its first generation of machine learning models five years ago and has been fine-tuning its efforts ever since. Back then, the network started with card accounts where fraudulent activity occurred, and where the company was able to trace back the historic transactions.
“It was through these data efforts and our machine learning models that we were really able to determine the most intersected merchants and locate a common point of purchase, which was more likely than not the originating source of the data compromise,” he said.
Now, the company is leaning more heavily on machine learning to enhance its existing fraud monitoring and detection — using natural language processing to analyze patterns in the dispute texts and notes to find some potentially illegal or deceptive merchants.
“We’ve really come a long way and continue to invest in machine learning and AI technology to help us in managing against fraud,” as it [the technology] shares insights with network partners, Stucchio said.
The company’s account management tool, he said, helps to exchange information with other entities, informing acquirers so that they may correct merchant loopholes. Through the company’s High Brand Risk program, the network works with acquirers to identify, register and manage merchants that engage in riskier business operations, such as online gambling or cryptocurrency.
“We are working to enhance our existing fraud-reporting platforms so that we can do a better and more seamless job of exchanging information with issuers about fraud incidents and provide a more fulsome picture of the fraud landscape,” he said. In doing so, he said, the returns on investment go well beyond the confines of dollars and cents — they extend to brands’ reputations.
We’re always on the lookout for opportunities to partner with innovators and disruptors.
Learn More