Finance Teams Finally Have Proof AI Fraud Tools Pay Off

 | 

Highlights

Seventy-seven percent of firms using AI-powered fraud defenses say they get better results than they did with manual methods, according to PYMNTS Intelligence’s “Prevention First: Building a Smarter Defense Against Payments Fraud,” produced in collaboration with Bottomline.

The data shows that 59% of firms using AI tools say they stop at least 90% of fraud attempts before any money leaves the business. Among firms without AI fraud tools, 32% report the same level of prevention.

Companies can measure whether more attempted fraud is being intercepted before a payment becomes a loss.

Prevention Before the Payment

Timing is critical because many of the fraud schemes companies encounter are designed to manipulate a legitimate payment process.

Half of firms say AI-written emails impersonating vendors or suppliers are the AI-enabled fraud tactic they encounter most often, while 88% encounter such emails at least a few times a year. Nearly one-third, 32%, encounter business email compromise at least monthly.

Stopping those attacks can depend on what happens between receiving a payment instruction and releasing the funds.

The report describes firms using real-time risk scoring, automated document verification, email scanning, synthetic identity detection and behavioral biometrics as parts of their defenses. Among AI adopters, 83% use real-time predictive risk scoring and 82% use automated document verification. Seventy-eight percent scan incoming emails and texts for signs of AI generation.

Those controls address different points in a transaction. An email can be examined for signs of manipulation, documents can be authenticated, a transaction can be scored as it occurs and payment details can be checked against known information before funds are released.

The data also shows that companies generally aren’t deploying those capabilities one at a time. Forty-two percent of all firms use three or more AI fraud tools, while just 1% use one or two. Fifty-seven percent use none.

Examples in the report show AI detection working with account validation, anomaly monitoring and trusted supplier information. In one case, machine learning flagged the payee account on a suspicious wire request. Another firm’s anomaly monitoring caught an attempt to reroute an ACH payment after the instructions diverged from the vendor’s normal pattern.

The examples illustrate how AI can become part of the payment process itself. Detecting an unusual request is useful, but preventing the loss requires controls capable of acting on the signal before the transaction is completed.

Payment method can affect the equation as well. Thirty-five percent of firms report experiencing check fraud, and 48% say checks account for at least 10% of their outbound fraud losses. Fifteen percent attribute more than one-quarter of those losses to checks.

More companies appear ready to test the results for themselves.

Fifty-eight percent of firms that don’t currently use AI fraud detection say they plan to implement it within the next 12 months.

Their starting point is markedly different from that of existing adopters. Fifty-five percent of non-users say their manual reviews already work well enough, and they are 89% more likely than AI users to hold that view.

At PYMNTS Intelligence, we work with businesses to uncover insights that fuel intelligent, data-driven discussions on changing customer expectations, a more connected economy and the strategic shifts necessary to achieve outcomes. With rigorous research methodologies and unwavering commitment to objective quality, we offer trusted data to grow your business. As our partner, you’ll have access to our diverse team of PhDs, researchers, data analysts, number crunchers, subject matter veterans and editorial experts.