Visa/Featurespace Details How Fraudsters Use AI to Mass-Produce Checks
- Visa reports checks drive nearly one-third of U.S. fraud losses, outpacing digital
- Criminals now blend check washing and social engineering with faster payment scams
- AI image forensics plus behavioral analytics help banks contain cross-channel check fraud
In a new PYMNTS post, Visa’s Michele Herron warns that even as digital payments accelerate, old-fashioned checks remain a massive liability — driving nearly a third of U.S. fraud losses and proving 31 times more likely to be fraudulent than real-time payments.
Click the image above to watch the video.
Ms. Herron provides insights on how fraudsters take advantage of the time lag between deposit and final settlement and details some of the tactics utilized.
Fraudsters use a range of methods, including counterfeit checks, forged signatures, and altered stolen checks, with check washing standing out as a prominent example: legitimate checks are intercepted, stripped of original ink, and rewritten. At the same time, many schemes rely on social engineering rather than document alteration, tricking victims into writing real checks under false pretenses, such as fake job offers or fabricated financial gains. As a result, check fraud spans both document manipulation and human deception, making it increasingly difficult to detect and prevent.
How AI Is Streamlining Check Fraud for Criminals
A key point noted by Ms. Herron is the fact that AI has enabled fraudsters to manipulate check images and mass-produce counterfeit checks at scale -- with remarkable accuracy.
“In the past it was painstaking to get the signature right and copy handwriting,” Herron said. “Now with AI, you can do this at scale...
...AI can look at nuances in ink patterns, the way numbers are formed, and how space is used,” she said. “It can reach nearly 100% precision.”
Because of advancements in generative AI, fraudsters can leverage the technologies to perform quick check image alterations and create hundreds of counterfeits within mere seconds. We warned readers back in 2023 in a post about generative AI as a major tool for fraudsters to create realistic counterfeit checks, noting that "It doesn't take too much imagination to see how fraudsters could use this readily available technology to quickly and efficiently create incredibly realistic checks."
This was confirmed in a recent post, where Uri Rivner, CEO and Co-Founder of Refine Intelligence, shared a shocking video showing just how easy it is for generative AI to assist fraudsters:
How Visa/Featurespace is Fighting Back with AI
A key point noted by Ms. Herron is the fact that AI has enabled fraudsters to manipulate check images and mass-produce counterfeit checks at scale -- with remarkable accuracy.
Ms. Herron details different layers of AI technologies utilized by Visa/Featurespace to detect counterfeits, forgeries, and alterations, explaining that technologies are needed to both assess the physical characteristics of checks and broader transactional signals, using highly accurate image forensics to identify issues such as bleach stains and subtle inconsistencies that may indicate a check has been altered or is counterfeit.
However, image forensics is only one layer of defense. Herron highlighted the role of adaptive behavioral analytics, enabled through Featurespace, which builds dynamic profiles of customer behavior to detect anomalies in account activity, payment patterns, and usage. By continuously learning what “normal” looks like, these models can more effectively flag suspicious transactions while allowing legitimate ones to proceed, helping institutions strike a balance between fraud prevention and a seamless customer experience.
As fraud techniques and tools continue to evolve, it is increasingly vital for financial institutions to employ a robust, multi-layered technology framework for on-us and deposit fraud detection to keep up with the fraudsters.