Artificial Intelligence for Audit, Forensic Accounting, and Valuation

CHAPTER 12Intelligent Automation of Fraud Detection

WE COVER AUDIT RELATED ERRORS AND FRAUD throughout this book; however, this chapter is dedicated to intelligent financial fraud detection and investigation (IFFDI) using artificial intelligence (AI). The internal controls environment serves as the firewall to protect against fraud. Prevention is different than detection and detection is different than investigation. Detection implies active search to discover, find, uncover, or expose fraud. Investigation begins once a reasonable degree of suspicion develops about the likelihood that a fraud is taking place. In this chapter we will cover both detection and investigation.

When deploying your fraud detection capabilities, you will address questions such as

• How to allocate capabilities between prevention, detection, and investigation? What technologies to deploy for each?
• When and for what to use deterministic versus stochastic technologies?
• What methods, models, or techniques should you use?
• How do you design and architect your IFFDI system?
• How to protect your firm from AI‐centric fraud?

DETECTING FRAUD

Fraud happens when material misrepresentations of facts are made intentionally and they lead to loss for one or more victims. The Association of Certified Fraud Examiners (ACFE) classifies fraud against a firm into two types: internal and external (ACFE, n.d.). Internal fraud, also known as occupational fraud, is defined by ACFE as “the use of one's occupation for personal ...