EY, one of the leading global accounting firms, has recently implemented artificial intelligence (AI) in its audit processes to detect fraudulent activities. Initial results have shown promising outcomes, with AI identifying suspicious activities in two of the first ten companies evaluated. While EY touts the potential of AI in auditing, the industry remains divided on its reliability for fraud detection.
AI in Audit: Detecting fraud
Ernst & Young (EY), a prominent player in auditing and accounting, has ventured into artificial intelligence to enhance its audit processes. By harnessing the power of AI, EY aimed to improve the accuracy and efficiency of its auditing procedures.
EY’s foray into AI-driven auditing began in 2018, intending to revolutionize how the company detects fraudulent activities. In the initial phase of this endeavour, EY used AI to assess ten different companies. Remarkably, the AI system identified suspicious activities in two of these companies, subsequently confirmed as fraud by the clients.
Kath Barrow, EY’s UK and Ireland assurance managing partner, expressed optimism about the effectiveness of the AI system. While specific details about the AI software and the nature of the detected frauds remain undisclosed, Barrow’s statement suggests that EY sees significant potential in incorporating AI into its auditing processes.
The birth of Helix GLAD
One of the key figures behind EY’s successful integration of AI into auditing is Naoto Ichihara, an assurance partner for Ernst & Young ShinNihon LLC in the Tokyo office. With a background in programming and a passion for developing models and systems for audit, Ichihara was inspired to explore the application of machine learning in accounting data analysis.
Ichihara’s extensive research into existing academic papers and algorithms led him to a groundbreaking realization: there was a more effective way to detect anomalies through machine learning. Driven by this vision, Ichihara embarked on the journey to create an AI solution capable of sensing irregularities within vast databases of financial information.
His innovative technology became the first of its kind in auditing, ultimately leading to a patent for this groundbreaking solution. Named EY Helix GL Anomaly Detector (Helix GLAD), it marked a significant milestone in AI-driven auditing.
Gaining trust in AI
While AI has the potential to analyze extensive datasets rapidly, the accounting industry remains divided on its reliability for detecting the myriad forms of potential fraud. To address scepticism, EY’s assurance team conducted rigorous testing of Helix GLAD against a dataset containing pre-determined fraudulent journal entries.
As the algorithm consistently and accurately uncovered these fraudulent entries, auditors began to gain confidence in the potential of Helix GLAD to enhance auditing accuracy. However, a critical element was still missing: auditors lacked insight into why the AI system detected specific anomalies. This knowledge was vital for assessing the validity and impact of the flagged entries.
Bridging the gap: Data analytics for transparency
Recognizing the need for transparency and understanding in auditing, EY’s team devised a solution that leveraged data analytics. This solution aimed to create visual maps of flagged entries, giving auditors insights into the reasons behind the AI algorithm’s detections.
These visual representations enabled auditors to evaluate the flagged entries comprehensively, fostering trust in the algorithm’s detection methods. This transformation marked a significant step forward in improving the accuracy and efficiency of auditing processes while reducing the risk of financial irregularities going undetected.
Benefits and challenges of AI in audit fraud detection
Integrating AI into audit fraud detection processes offers numerous benefits for large accounting firms like EY. AI algorithms possess the capacity to analyze vast volumes of data in a fraction of the time it would take a human auditor. This efficiency allows auditors to focus on interpreting results rather than laboriously reviewing data. Moreover, AI models are not susceptible to human bias or fatigue, consistently applying predefined rules and criteria to identify anomalies.
This objective and reliable approach reduces the risk of overlooking suspicious transactions due to human error or oversight, thereby enhancing the effectiveness of fraud detection.
However, implementing AI in audit fraud detection comes with challenges. Integrating AI technology into existing auditing systems and workflows is a significant challenge. Accounting firms must ensure that AI algorithms seamlessly align with their infrastructure and processes, necessitating careful planning, training, and collaboration between auditors and AI specialists.
Another challenge revolves around the continuous monitoring and updating of AI algorithms. As fraudsters adapt and evolve their techniques, AI algorithms must remain adaptable to detect new patterns and anomalies. Collaboration between auditors and developers is crucial for refining and updating algorithms to stay ahead of emerging threats.
The future of AI in audit and regulatory considerations
The adoption of AI in audit fraud detection has the potential to enhance audit quality and efficiency. However, regulatory bodies will play a pivotal role in determining the extent to which accountants can rely on AI during the audit process.
Jason Bradley, head of assurance technology for the UK’s Financial Reporting Council, acknowledges that AI presents opportunities for improving audit quality and efficiency if used appropriately. Regulatory decisions likely hinge on accountants’ ability to critically evaluate and critique AI systems.
Additionally, the issue of data ownership poses a challenge. Companies may view their detailed financial data as proprietary information, making using such private data to train AI systems for auditing other entities complex.
EY’s successful implementation of AI in audit fraud detection highlights the potential benefits of AI in auditing. While challenges persist, the transparency and efficiency brought about by AI-driven solutions can reshape how auditors detect and address fraudulent activities in the future. The industry, regulators, and auditors must collectively navigate these opportunities and challenges as AI continues evolving in auditing practices.
