In the ongoing effort to combat check fraud, a recent advancement could help in addressing a previously neglected aspect of risk.

ParaScript, an organization that specializes in AI-driven document processing, has unveiled a new feature designed to detect and interpret handwritten or stamped endorsements on the back of checks. This update to Parascript’s check recognition solution, CheckXpert.AI, allows the system to identify phrases such as “For mobile deposit only” or “Deposit only to account of payee.” The technology automatically detects the check’s orientation and matches the text against a customizable list of authorized phrases, flagging any endorsements that seem missing, suspicious, or unauthorized.

This enhancement aids in ensuring that checks are deposited through the correct channels and minimizes fraud risks. However, it’s crucial to understand that the tool does not verify handwriting authenticity or match signatures to those on file—capacities provided by some other investigative tools.

A Practical Enhancement

The automation of a process known for causing difficulties in remote deposits—endorsement verification—is beneficial.

“Most fraud solutions concentrate on signatures or altered fields, but this fills a smaller gap by ensuring the back of the check contains the correct information,” remarked Jennifer Pitt, Senior Analyst of Fraud Management at Javelin Strategy & Research. “The tool can flag issues in real time, preventing them from being accepted before review, rather than catching them later during back-office processing.”

“It’s a practical improvement, albeit not a significant leap in handwriting analysis,” she continued. “For banks handling high volumes of deposits or stringent compliance requirements, it reduces the need for manual scrutiny and ensures more consistent adherence to basic control measures.

AI Innovations

AI has become an essential component in the battle against check fraud. The federal government remains a significant user of paper checks, with 23% of benefit recipients receiving payments via this method. They have employed AI to detect fraudulent activity, achieving highly successful results. According to CNN, machine learning technology assisted the Treasury in recovering $1 billion from check fraud in fiscal 2024—a figure nearly three times that of the previous year.

Despite issuing approximately 700,000 reports of check fraud annually, many organizations still do not prioritize this form of crime. According to Javelin’s survey, only 22% of companies use check fraud detection solutions.