As fraud gets more sophisticated, credit unions turn to artificial intelligence

Register now

With scammers steadily improving their techniques, CO-OP Financial Services is starting to offer artificial intelligence to police transactions initiated by online, mobile and other connected devices.

"It's exciting, the payments industry has changed dramatically," said Stephanie Pike, product manager at CO-OP. "But fraud is changing just as quickly."

CO-OP, which provides shared branching and technology to a network of member credit unions, has embarked on an AI project that will initially focus on payments risk management. The credit union service organization is investing $20 million on the technology, as CO-OP says more traditional fraud fighting tools have not kept pace with innovation in crime.

The security phase of the project, which involves a partnership with Feedzai to co-develop AI applications, will be completed this year. It's designed to build on other fraud tools such as behavioral analysis. By applying machine learning, CO-OP hopes to improve tasks such as spotting suspicious payments.

Behavior analysis as fraud prevention has been around for years, but is getting fresh attention among financial institutions as a way to bolster marketing and security.

"Machine learning and artificial intelligence have become table stakes due to the growing sophistication of today's fraudsters," said Al Pascual, a senior vice president and research director for Javelin Strategy & Research. "Fraudsters are diversifying their schemes and leveraging automation in such a way that most institutions, which are overly reliant on manual prevention and detection of fraud, have become increasingly vulnerable."

Large payment companies such as PayPal and Mastercard have also deployed AI to make security less intrusive. As AI "learns" more about consumer behavior, it should produce fewer false positives when flagging transactions.

"There is a lot of substance behind all the buzz that we’re seeing around the use of machine learning for fraud prevention," said Julie Conroy, research director for Aite, adding machine learning enables models to learn on an iterative basis and, therefore is proving quite effective at enhancing fraud mitigation efforts. "The success is such that those that do not invest in this technology risk being left behind, as their competition that have embraced it are able to provide superior customer experiences while also detecting more fraud."

Though AI's use in this area is still early, CO-OP sees AI as a dynamic innovation that can enhance existing IT strategy in resource management, security, authentication and digital transactions.

"This doesn't mean that machine learning can run on its own," Pike said. "There's human input. There's so many things feeding into it that we can go across the entire enterprise to get a new view of the customer."

Behavior analysis has also traditionally served marketing, making AI a potential fit for customer service and cross-selling, a potential that also interests CO-OP. Beyond security, CO-OP didn't reveal specific use cases for AI, but one of the benefits of the technology is there doesn't necessarily have to be—the improved analytics that machine learning can provide produces an evolving view of customer preferences and tendencies such as how and where they are paying, a mix of channels and form factors that can provide predictive analysis for product development.

As consumers use the credit union's digital systems, shared branch network or make payments with CU-issued cards, an increasing amount of data is collected that can inform myriad tasks from marketing to customer service to product research and development, according to Pike.

"We're able to take into account more than just what transactions they have on their debit card," Pike said. "It can combine with branch visits, or payments drawn from other accounts or other activities."

For reprint and licensing requests for this article, click here.
Artificial intelligence Venture funding Venture capital Fraud detection