The Role of the External Auditor in Detecting Tax-Evasion Risk Patterns Using Artificial Intelligence Algorithms in Light of Revised ISA 240: Evidence from Iraqi Listed Companies

Vol. 21, N°2 (2026)

The Role of the External Auditor in Detecting Tax-Evasion Risk Patterns Using Artificial Intelligence Algorithms in Light of Revised ISA 240: Evidence from Iraqi Listed Companies

Nadia Talib Salman, Fatimah Fezea Hadab and Lina Garbeet Krikor

Résumé

The amended International Standard on Auditing (ISA) 240 places a greater focus on professional skepticism and on technology-based procedures in relation to fraud-risk assessment, but there is less research on the relationship between artificial-intelligence-based detection and observable audit characteristics in emerging markets. In this study, three unsupervised methods namely Isolation Forest, Local Outlier Factor and One-Class Support Vector Machine are applied to detect the anomalies in the financial and tax attributes of non-financial companies in Iraq stock exchange. The sample of 153 observations from 28 companies during 2015-2024 is verified from an initial sample of 340 firm-year observations. The ensemble outclassed the remaining classifiers at a 12% contamination rate, identifying 16 tax anomaly years whilst One-Class SVM performed best with an F1 score of 0.300. While the rates of modified opinion, auditor change, state audit and audit report lag were descriptively higher, Firth did not find these rates to be significantly independently associated with any of the logistic regressions controlling for firm size, leverage and return on assets. A robustness model of the firm size was consistent with a significant state-audit relationship; however, this relationship was not apparent for the main specification. By using contamination rates of 10%, 12% and 15%, 12 stable anomalies were identified in sensitivity analysis. The findings overall justify the use of AI in addition to the judgment of external auditors as per revised ISA 240, not in place of it.

Mots clés:

External audit, ISA 240, Audit report lag, Tax anomaly detection, Artificial intelligence, Unsupervised machine learning, Firth logistic regression

Comment citer cet article

Nadia Talib Salman, Fatimah Fezea Hadab and Lina Garbeet Krikor (2026), " The Role of the External Auditor in Detecting Tax-Evasion Risk Patterns Using Artificial Intelligence Algorithms in Light of Revised ISA 240: Evidence from Iraqi Listed Companies", International Journal of Management And Technologies, Vol.21, N°2, pp: 52-69. http://www.ijmat.tn/pages/articles/31/186