AI-Augmented Auditing: Enhancing Accuracy, Coverage, and Predictive Testing in Assurance

Abstract

The rapid digitalization of business environments and the exponential growth of structured and unstructured financial data have exposed significant limitations in traditional audit methodologies, particularly those relying on sampling, manual review, and periodic evaluation. Artificial intelligence (AI) introduces a transformative shift in auditing by enabling full-population analytics, enhancing anomaly detection accuracy, and supporting predictive testing that anticipates risk before it materializes. Research demonstrates that AI-driven auditing tools including machine learning (ML), natural language processing (NLP), computer vision, and process mining, can analyze 100% of transactional data, uncover patterns invisible to conventional sampling, and reduce detection time for control failures; [5], [6]. ML algorithms have shown accuracy rates exceeding 90% in identifying high -risk or fraudulent transactions when trained on high-quality datasets [3], while NLP accelerates document analysis and improves the identification of inconsistencies in disclosures and contracts; [3]. Process mining similarly enhances coverage by identifying more control deviations compared to manual walkthroughs [10]. Overall, AI-augmented auditing represents a paradigm shift from retrospective, sample-based assessments toward comprehensive, real-time, and predictive assurance models. Rather than displacing auditors, AI elevates their role, enhancing professional judgment, strengthening assurance reliability, and enabling deeper risk insights in increasingly complex financial ecosystems.