Artificial intelligence and statutory audit: Machine Learning in the service of fraud detection

Abstract

The rise of Artificial Intelligence (AI), and particularly Machine Learning (ML), represents a major transformation for statutory audit, disrupting traditional methods and raising crucial questions about improving fraud detection. This evolution, rich in promises of efficiency yet full of challenges regarding technological integration, requires a reassessment of the role, skills, and approaches of statutory auditors in the analysis of financial statements. To address this issue, this article draws on a narrative and conceptual literature review and mobilizes various theories (agency, signaling, risk, etc.) in order to build a solid theoretical understanding of the impact of ML on fraud detection. The analysis reveals that Machine Learning algorithms significantly enhance the speed and accuracy of anomaly identification, reduce false positives and negatives, and strengthen the proactive assessment of fraud risks. However, this integration may also require an adaptation of auditors’ skills and raise concerns about the interpretation of algorithmic models. This theoretical contribution highlights the synergies between technological innovation and the imperative of integrity in statutory audit, while suggesting avenues to sustainably adapt this strategic function to the digital era, ensuring trust and reliability in financial information.

Classification JEL : M42, M4

Paper type : Theoretical Research