As artificial intelligence continues to affect our lives and the business that we transact, its evolution has provided a new opening for those who commit fraud. According to industry estimates, fraud powered by AI is expected to reach $10.5 trillion by 2025.
As organizations seek new ways to combat this fraud, they must be careful not to alienate their customers in the process. A multi-layered fraud prevention framework bolstered by advanced AI and machine learning-based technology achieves these objectives by proactively mitigating risk while minimizing fraud losses. In a recent PaymentsJournal webinar, Max Spivakovsky, Senior Director of Global Payments Risk Management for Galileo Financial Technologies, spoke with Kevin Libby, Fraud and Security Analyst for Javelin Strategy & Research, about how fraud risks have evolved in an age of AI and what organizations are doing to combat emerging risks
Where AI Is Headed
The evolution of AI and machine learning opened the doors for advanced modeling capabilities, advanced pattern recognition, and behavior analysis. It has generated an adaptive learning of customer activity and behavior. FIs are also increasingly using chatbots with intelligent digital assistants (IDAs) that interact with customers in real-time to address emerging fraud risks. This is another step in the evolution of AI as it relates to fraud.
The natural language processing capabilities that it opened—and the ability to address those models in a faster way—is a huge leap in in the fraud controls available. As for the impact on customers, Spivakovsky said AI-powered fraud mitigation allows financial institutions to enhance their overall fraud and risk analysis approach.
“Model creation is automated and recursively learned from previous experiences, such that exceptions requiring manual review become less and less common over time,” Spivakovsky said. “That’s a huge win for commercial enterprises and for financial institutions, in that it frees up human capital that would otherwise be tied up with those manual reviews. That, in turn, allows them to utilize their workforce more efficiently and to stretch departmental resources a bit further than they otherwise could.”
Automation has been particularly helpful in helping financial institutions get through the mountain of suspicious-activity reports, for example, that they are required to file every month. AI allows for the creation of more complex models because it is capable of creating rules or models that digest larger number of testable parameters than manually created rules-based systems ever could.
Taking a Proactive Approach
A reactive approach can’t stay ahead of payment fraud trends. To stay relevant in the industry, financial institutions must deploy proactive approaches.
A proactive approach enables the detection of anomalies faster than manual, reactive fraud and risk analysis.
“The link analysis and accuracy of the models make the proactive approach so much more accurate,” Spivakovsky said. “Some of the examples available on the market right now are able to notify the financial institutions or the customers that they might be subject to potential fraudulent activity. For them to save the financial means, we can either replace the card or even restrict some of the customer spend. Being more reactive means we keep our hands on the pulse all the time in terms of model accuracy.”
Proactive fraud prevention systems are set up not only to determine which payment cards are already experiencing fraud but also to determine the potential number of cards at risk to experience fraud because of that compromise. When AI-powered fraud detection tools are used to make those types of predictions, the technology relies on a wealth of data and learns from previous fraud incidents. AI tools can better pinpoint the scope of a potential compromise and proactively identify the accounts most at risk
Breaking Down the Silos
One of the biggest challenges in protecting multi-channel systems is that each channel provides its own set of testable parameters to identify fraudulent activity. Some channels have more robust data to scrutinize than others, and some don’t have access to much data at all when addressed in isolation. It’s critical to break down these silos and consolidate an organization’s efforts.
“In the old way of doing things, you had to create separate models for detecting and preventing fraud in each individual channel without really incorporating information you may have from the last time you interacted with a given user by a different channel,” Libby said. “Things tended to be segmented and isolated. One strength of AI-assisted decisioning is the ability for a program to incorporate data from across those various sources.”
The customer experience with using chatbots, and the ability for the client to complain in real time about a specific incident, allows organizations to convert this input into actionable methodologies within the operational universe or within the first line of defense. This gives the existing models the ability to learn much more quickly.
Today, financial institutions have more customer data than ever, including from incidents being flagged in real-time via customers using chatbots. The implementation of AI and machine learning models allows organizations to gain actionable insights from all this data to create quicker line of defense to proactively stay one step ahead of fast-moving fraudsters.
“I usually talk about it in terms of a digital arms race: the criminals and the cybersecurity and fraud professionals trying to stay one step ahead of each other all the way,” Spivakovsky said. “The difference between what we’ve seen recently and what we’re going to see in the near future is that given natural language models, the pace of trying to outdo one another is only going to increase. We’re going to be playing catch-up for a while, but hopefully in the end we still figure out how to stay that one step ahead.”
In today’s digital landscape, AI and machine learning-based fraud prevention technologies stand as essential allies for banks and fintech companies. By actively identifying and thwarting fraudulent activities, these advanced systems not only save significant costs incurred from fraud losses but also shield the reputation of financial entities from potential harm. And their proactive approach not only bolsters security but also instills confidence among customers, ensuring a resilient and trusted financial ecosystem.
