Alteration fraud involves criminals “check washing’ or “check scraping” a negotiable check by removing or replacing the payee or check amount. Fraudsters rewrite the check for a new recipient with a larger sum before cashing the check. In many instances these fraudulent checks can be altered for hundreds or thousands of dollars more than what was originally written. Unlike counterfeit checks that are 100% fake, alteration fraud uses valid authorized checks, making detecting this fraud type more complex due to the combination of valid and invalid content.
Combating this growing problem requires embedded technology solutions with image analysis depth that leverage newer artificial intelligence (AI) tools and techniques. In addition to the traditional evaluation of structured data, solutions that use “deep learning” models, also known as artificial neural networks (ANNs), can evaluate checks for subtle differences in handwriting style or fonts. Solutions that incorporate AI models trained on large, diverse data sets of handwriting samples enable them to learn and recognize a wide range of handwriting styles.
Sophisticated software analyzes the check image to ensure the handwriting matches across check fields. Identification of handwriting variations between the amount or the payee field and other fields may indicate that alteration has occurred. Additionally, the analysis from these types of solutions compares handwriting to valid reference checks previously provided to the system. If the handwriting styles differ from previous images, the system marks the checks as suspicious.
Many systems often lack the ability to compensate for poor image quality. Certain solutions can apply AI enhancement techniques to clean up artifacts caused by noise, color, brightness, skew, and other problems caused by handling and scanning issues. These AI-driven learning techniques help maintain high fraud detection rates while reducing false positives that difficult-to-manage processes might otherwise introduce.
For more information download our interview: Fraud & AML Solution Overview
