Examining the role of Gen AI in combating RTP fraud

With significant current and expected growth in the adoption of real-time payments (RTP), RTP fraud – particularly, authorized push payment (APP) fraud – is also set to sizably increase. For financial institutions and RTP networks, this poses both an opportunity and a challenge. Let’s look at the role generative AI can play in developing more effective fraud detection and prevention systems that can help detect and reduce APP fraud.

2023: A banner year for RTP and payment fraud

2023 was an inflection year for RTPs. Following the successful launch of FedNow, the third U.S. real-time payments rail, and the rapidly expanding set of participating banks, ACI projects RTP transaction volumes are expected to reach $4,698.7 billion in 2027–a CAGR of 41.3% over 2022-27.

Not surprisingly, fraud is also hitting new highs. According to the U.S. Federal Trade Commission, U.S. consumers lost at least $8.8 billion to fraud in 2022, a 44% increase over 2021. Because RTPs are instantaneous and irreversible, they raise the stakes for fraud even higher. Especially troubling is the rise in APP fraud, which ACI predicts will exceed $3.03 billion in the U.S. by 2027, up from $1.94 billion in 2022. The most prevalent were imposter scams, in which victims are tricked into giving money to fraudulent individuals or companies posing as trustworthy sources.

Examining the role of Gen AI in combating RTP fraud

With its ability to:

• analyze significant volumes of fraudulent transaction data to detect patterns and trigger events,
• generate significant volumes of synthetic data based on behavioral analysis, and
• leverage an ever-expanding dataset to rapidly learn from to enable its algorithms to mature faster,

Gen AI can develop predictive models to generate accurate, up-to-the-minute insights and flag suspicious activity in real time.

Given its flexibility for capability enhancement, Gen AI holds sizeable promise as a weapon against RTP fraud. Another important differentiator is accuracy. Case in point, false positives are known to drive customer churn: Two-thirds of cardholders who experienced a false-positive decline during an e-commerce transaction reduced or stopped their patronage of the merchant. Gen AI-trained fraud detection tools will have the ability to differentiate fraudulent transactions from legitimate purchases to a high degree of accuracy, thereby reducing the number of false positives and dissatisfaction outcomes. Banks and FIs need to be vigilant and ensure customer PII data is protected as they build and grow the data sets to mature their Gen AI offering.

Emerging innovations

We’re seeing aggressive investment in generative AI solutions for the payment industry and the innovation resulting from those investments. For example, DataVisor’s AI Co-Pilot enhances financial institutions’ ability to detect fraud while reducing false positives and minimizing user friction, all in real time. The solution uses Gen AI to not only automate rule-tuning and feature-script generation but also generate natural language-based rule descriptions that humans can easily comprehend.

While Gen AI’s true impact in aiding RTP fraud prevention remains as yet unknown, I’m enthused by the art of the possible that Gen AI presents. I’ll continue to share my viewpoint on this evolving topic and look forward to exploring the potential value it can deliver for our payments clients.