AI and Fraud Detection: Insights from ACFE Research Institute Grant Recipient

Fraud detection is a critical yet challenging aspect of auditing and financial oversight. With the support of the ACFE Research Institute, researchers are working to uncover insights that can enhance fraud prevention and risk assessment.

In this interview, we speak with Ivy Munoko, an assistant professor at the Fisher School of Accounting, Warrington College of Business, University of Florida, about her research under the ACFE data access grant. Alongside her colleagues, she is investigating the characteristics that influence auditors' ability to detect fraud, as well as the potential of artificial intelligence (AI) in corporate fraud detection. Through her work, she aims to provide valuable insights for regulators, auditors and fraud examiners navigating an increasingly complex landscape. 

What is the focus of your current research under the ACFE Research Institute grant? 

Our research, supported by the ACFE data access grant in 2024, examines the characteristics associated with the successful detection of fraud through audit mechanisms. The ACFE estimates that external auditors detect only 3% of frauds. Furthermore, only 9% of frauds discovered by other parties (e.g., management and internal auditors) are brought to the attention of external auditors. Our research examines what audit client and environmental factors are associated with auditors' successful fraud detection, which would be informative to regulators and the audit profession. I am conducting this research with my colleagues, Syrena Shirley, an assistant professor at Columbia Business School, and Triza Nganga, Ph.D. candidate at the University of Texas at Arlington. 

A second research project, supported by the ACFE Research Institute grant in 2023, evaluates the net benefit of using AI (large language models and machine learning models) in detecting corporate fraud by analyzing a large dataset of corporate managers' discussions on their firm's financial performance. The research publication is currently under journal review. I conducted this research with Helen Brown-Liburd and Soohyun Cho, both associate professors at Rutgers Business School. 

How do you believe your research will contribute to the field of fraud examination and prevention? 

The findings from our studies highlight that detecting fraud risk factors and events is complex, requiring a multipronged approach that considers the organizational context and temporal changes in corporate communication. The studies also highlight the complexity of detecting fraud risk and demonstrate the potential to enhance auditors' detection of fraud risk by incorporating advanced technology. 

What are some of the biggest challenges you have faced in your research? 

It is hard to obtain a comprehensive fraud dataset to address our research questions effectively. As a research team, we are very grateful to the ACFE for the rigorous and consistent data collection over two decades that enables us to understand this phenomenon better. Also, obtaining access to professionals to participate in our studies is often challenging. Through the online ACFE Community and during the ACFE Global Fraud Conference, we met professionals who participated in our studies and provided us with valuable insights.

My connection to the ACFE is one of my most valued professional relationships, significantly informing my research. I am especially thankful to Kip Holderness, professor at West Virginia University and the ACFE Research Institute director of research; Mason Wilder, research director at the ACFE; Andi McNeal, chief training officer at the ACFE; and Yasmin Hoffmann, educational production director at the ACFE, who have all been instrumental to our research. 

What do you believe are the practical implications of your research for fraud examiners? 

Our research findings from studies supported by the ACFE point to the difficulty of detecting fraud, specifically in large organizations. Our research shows that analyzing corporate communication provides useful signals of fraud. However, it may be difficult for auditors to review corporate communication manually due to time and resource constraints and the typical volume of corporate communication. Our study documents the capability of language models to automate preliminary risk assessments using public and private corporate communication, which are typically accessible to auditors.  

What advice would you give fraud examiners and researchers interested in applying for an ACFE Research Institute grant? 

Start with a research question that is important and interesting to you. Share this research with your trusted colleagues to get their feedback. This will help you refine your research question and find potential collaborators. You may be tempted to keep finetuning your idea until it's perfect, but as the deadline approaches, do not hesitate to submit your proposal, imperfect as it may be. Ultimately, I believe an informative and practical research idea is the key to obtaining grants.