RANCHO CUCAMONGA, Calif.—The human eye is no longer a reasonable defense against cyberthieves, asserts CO-OP Financial Services, which expects many more credit unions next year will invest in artificial intelligence-backed fraud fighting tools.
The U.S. migration to EMV chip cards has spurred a steady increase in card-not-present fraud over the past few years, and as fraudsters have made the move to digital environments, the cost and complexity of fraud detection and prevention has grown exponentially, pointed out Fotis Konstantinidis, senior vice president, fraud products for CO-OP.
“Human analysts operating rules-based systems, coupled with simple machine learning (ML) techniques—single-layer neural networks—were largely sufficient in the past, spotting and stopping most fraud before it could cause too much damage,” explained Konstantinidis. “However, with new tools and technologies at their disposal, criminals are moving faster, they are leaner, and they have developed automated ways to deploy extremely well-crafted and highly targeted schemes. The fact is the human eye can no longer keep up with the pace and sophistication of fraud coming from a number of new digital channels. The good guys are not the only ones on a digital transformation journey.”
In the coming year, more credit unions will invest in similar technologies to level the playing field, predicted Konstantinidis, with artificial intelligence and advanced ML at the top of the solutions heap.
“Well-suited to beat the bad guys at their own games, these technologies have become somewhat democratized, meaning you don’t necessarily need deep pockets and a team of data scientists to deploy and test a number of AI techniques,” said Konstantinidis. “The beauty of AI and ML is the learning, and not just for the machines. Nearly every AI platform on the market today functions with a human in the loop. So, there is knowledge flowing back and forth, from human to machine and vice versa.”
Interpretability is the key element of advanced AI for fraud and other AI-focused use cases, where the machine provides the precise reasoning behind decision making, explained Konstantinidis.
“It is very important for AI to allow humans to understand not only what goes in and comes out, but also what happens inside the AI box—why it made the decision it made or gave the recommendation it gave,” he said. “It is my belief that human fraud analysts will not be replaced by AI, at least not for the foreseeable future. Humans will be essential to AI’s deployment and to its evolution, especially in this most people-centric of industries.”
While saving dollars from stopped fraud is of paramount importance to credit unions, Konstantinidis said CUs are also excited about AI and ML’s potential to improve the overall member experience as they transact.
“The technology spots complex behavioral patterns and makes assumptions—along with a human in the loop—that significantly lower false positives and eliminate the payments friction that modern members often experience,” said Konstantinidis. “The investment will need to be somewhat collaborative, and that’s because AI and ML are at their best when fed a large and consistent diet of diversified data. And that appetite for data only gets bigger as the machines learn.”
The Greatest Benefit
Credit unions will benefit most from AI and ML platforms that pull high-quality, super-rich data from multiple channels, geographies, payment types and fraud investigation databases, Konstantinidis said.
“Because the CO-OP ecosystem is comprised of connections to many of these sources, we have endeavored to help satisfy the industry’s growing demand for AI and ML fraud prevention solutions,” he said. “That is precisely why CO-OP has invested millions in the development of COOPER, a suite of AI and ML solutions designed and customized specifically for credit unions. COOPER Fraud Analyzer is currently in beta with several U.S. credit unions. COOPER Fraud Score is set to launch in 2019.”
Understanding, predicting and explaining member behavior is at the core of what advanced AI provides, unlike more traditional approaches, emphasized Konstantinidis.
“Fraud is the primary use case for AI, but there are plenty of other ways to leverage the technology to personalize consumer experience and bring credit unions closer to their members,” he said. “I expect we will see many more use cases roll out in 2019 as the technology finds its place within the movement.”