Artificial intelligence and auditing: Legal and ethical implications for assurance providers

Abstract

Artificial intelligence (AI) is changing audit practice by expanding analytical coverage, automating routine procedures, and supporting more continuous risk assessment. These benefits, however, also raise questions about accountability, explainability, professional judgment, data protection, and legal responsibility. This study examines whether AI auditability and oversight design shape assurance professionals’ perceptions of ethical risk, legal exposure, and audit quality in Ghana. Using a quantitative design, the study combines a cross-sectional survey of 180 assurance professionals with a between-subjects vignette experiment. AI auditability was assessed in terms of data traceability, model documentation, explainability, governance controls, role clarity, and human-in-the-loop (HITL) procedures. The experiment compared manual-only auditing, AI-only auditing, and AI-assisted auditing with HITL oversight. AI auditability was negatively associated with perceived ethical risk (β = −0.456, p < .001) and positively associated with perceived audit quality (β = 0.424, p < .001). Perceived ethical risk was positively associated with perceived legal exposure (β = 0.536, p < .001) and negatively associated with perceived audit quality (β = −0.268, p < .001). Although AI auditability was negatively correlated with perceived legal exposure, its direct regression effect was not statistically significant after ethical risk was controlled (β = −0.092, p = .392). The experiment showed that AI-only oversight was associated with higher perceptions of ethical and legal risk than AI-assisted auditing with HITL oversight. Because the study measures professional perceptions rather than observed legal events or realized audit outcomes, the findings should be interpreted as evidence about perceived risk and quality. Overall, the results indicate that responsible AI adoption in auditing depends less on automation alone than on auditable governance, meaningful human oversight, and clear accountability.