Proactive Internal Fraud Detection Strategies Utilizing UX Data Based on Multiple Sensors

Abstract

The COVID-19 pandemic forced companies to change how they work, promoting a work style unbound by time or location, such as teleworking. While convenient, this remote approach has been linked to increased internal fraud risks due to alienation, perceived unfairness, and reduced compliance awareness. In 2023, Japan's Information-technology Promotion Agency (IPA) ranked "information leaks due to internal misconduct" as the fourth most critical information security threat. Internal fraud is categorized into carelessness/negligence (60%) and intentional misconduct (40%), with the latter costing 1.34 times more to address, leading to significant financial losses. This paper explores user experience (UX) environments utilizing sensors to collect non-cyber information, such as facial expressions, to counter intentional fraud. A theoretical analysis indicates that combining two sensor types—environmental and biometric—offers the most cost-effective solution, significantly improving fraud detection rates. While adding more sensor types enhances accuracy, cost-effectiveness declines beyond three types. Additionally, the detection of rationalization remains consistently low, highlighting the need for complementary methods like text and speech analysis or long-term behavioral monitoring. These findings underscore the effectiveness of UX-based non-cyber information and its potential as an innovative approach to mitigating intentional internal fraud.