Artificial intelligence (AI) is rapidly transforming accounting and audit functions through automation, anomaly detection, and predictive analytics. Simultaneously, it introduces new assurance and governance risks, particularly “automation bias”—the propensity of users to over-trust algorithmic recommendations and reduce professional skepticism. This paper examines how automation bias affects audit judgment, internal control effectiveness, and compliance readiness under the European Union Artificial Intelligence Act (European Union, 2024). Using a socio-technical risk model and control-mapping approach aligned with COSO Internal Control–Integrated Framework and ISA 315 (Revised 2019) (Committee of Sponsoring Organizations of the Treadway Commission COSO, 2013; International Auditing and Assurance Standards Board IAASB, 2019), the study proposes a practical governance blueprint for companies and audit firms operating in Western Balkan supply chains connected to EU markets. Findings emphasize that AI-enabled controls can increase coverage and timeliness, but may degrade control reliability without robust human oversight, explainability, monitoring, and model risk management. A compliance roadmap integrates AI Act obligations with audit evidence requirements and risk management standards (National Institute of Standards and Technology NIST, 2023; International Organization for Standardization ISO, 2023).
Harun Tahiri (Tue,) studied this question.