Algorithmic Bias and Prediction Errors in Machine Learning-Based Financial Auditing: Evidence and Challenges from Saudi Arabia

Abstract

Machine learning (ML) is increasingly transforming financial auditing by enabling continuous risk assessment, automated anomaly detection, and efficient examination of large transaction volumes. However, biased training data and uneven prediction errors may produce unfair audit decisions, expose particular organizational groups to disproportionate scrutiny, and weaken trust in algorithm-supported assurance. This study investigates the prevalence, sources, and consequences of algorithmic bias in ML-based financial auditing within Saudi Arabia. A three-year dataset comprising 285,000 anonymized financial transactions from 24 Saudi organizations across banking, insurance, retail, manufacturing, and public-service sectors was analyzed. The dataset contained 46 financial, behavioral, temporal, organizational, and audit-risk variables; 4.8% of transactions were verified as anomalous. Following cleaning, normalization, categorical encoding, class balancing, and a leakage-free 70:15:15 train-validation-test split, logistic regression, support vector machine, random forest, XGBoost, LightGBM, and artificial neural network models were evaluated. Predictive performance was assessed using accuracy, precision, recall, F1-score, area under the receiver operating characteristic curve (AUC), false-positive rate, and false-negative rate. Fairness was examined across organization size, sector, ownership type, and geographic region using demographic-parity difference, disparate-impact ratio, equal-opportunity difference, equalized odds, and subgroup calibration. XGBoost achieved the strongest baseline performance, recording 96.4% accuracy, 93.1% precision, 91.7% recall, a 92.4% F1-score, and an AUC of 0.982. Nevertheless, error analysis revealed substantial subgroup disparities: small and medium-sized enterprises experienced a 12.8% false-positive rate compared with 5.1% for large organizations, while non-central regions exhibited a 6.9-percentage-point higher false-negative rate. Historical audit-label imbalance, missing contextual variables, sectoral underrepresentation, and proxy features were the principal bias sources. A fairness-aware reweighting and threshold-calibration strategy reduced equal-opportunity difference from 0.143 to 0.052 and demographic-parity difference from 0.168 to 0.061, while accuracy declined only from 96.4% to 95.8%. The findings demonstrate that high aggregate accuracy does not guarantee equitable auditing outcomes. The study contributes Saudi-specific empirical evidence and proposes a governance framework integrating subgroup validation, explainable AI, human oversight, periodic bias audits, and regulatory accountability. These findings support Saudi Vision 2030 by guiding financial institutions, auditors, and regulators toward transparent, reliable, and responsible adoption of ML-enabled auditing across the Kingdom.

References

Ahmad, B., Majeed, M. K., Soomro, A. A., & Jillani, S. A. (2026, January). Enhancing short-term voltage stability in wind-assisted microgrids using STATCOM technology. In 2026 1st International Conference on Innovations in Information and Communication Technologies (IICT) (pp. 1–6). IEEE.

Ahmed, B., Jillani, S. A., Rafique, M., & Soomro, A. A. (2026, January). Design and monitoring of a tree-inspired vertical axis wind turbine for efficient urban wind utilization. In 2026 1st International Conference on Innovations in Information and Communication Technologies (IICT) (pp. 1–6). IEEE.

Al-Daoud, K. I., & Abu-AlSondos, I. A. (2025). Robust AI for financial fraud detection in the GCC: A hybrid framework for imbalance, drift, and adversarial threats. Journal of Theoretical and Applied Electronic Commerce Research, 20(2), Article 121. https://doi.org/10.3390/jtaer20020121

Al Fraidan, A. (2025). Procedural transparency and legal accountability to sustain AI-mediated language assessment in Saudi Arabia. SAGE Open, 15(4). https://doi.org/10.1177/21582440251396113

Al-Harbi, H. A., Al-Madailwi, S. N., & Al-Madailwi, R. H. (2026). Auditing in the modern era: The effect of using the AI tools on external auditors' ability of detecting fraud. Radiant Journal of Business & Sustainability, 1(1), 37–56.

Al-Omush, A., Almasarwah, A., & Al-Wreikat, A. (2025). Artificial intelligence in financial auditing: Redefining accuracy and transparency in assurance services. EDPACS, 70(6), 1–20. https://doi.org/10.1080/07366981.2025.2459490

Albaqshi, R. J., Najaei, H. A. A., Abdulmalik, H. G., Alzamaa, S. S. S., Alshahrani, S. O. H., Alshsari, M. M. M., & Najdi, M. S. (2025). Algorithmic fairness in clinical predictive models: A review of bias audits and mitigation strategies in epidemiological research. Saudi Journal of Medicine and Public Health, 2(2), 3073–3084.

Aldemir, C., & Üçma Uysal, T. (2025). Artificial intelligence for financial accountability and governance in the public sector: Strategic opportunities and challenges. Administrative Sciences, 15(2), Article 58. https://doi.org/10.3390/admsci15020058

Alharbi, R., & Abbadeni, N. (2026). Software unfairness detection in machine learning-based systems: A systematic mapping study. Software, 5(2), Article 18. https://doi.org/10.3390/software5020018

Alharbi, Y. (2023). Audit data analytics, the transformation within the audit profession: Perspectives from the Kingdom of Saudi Arabia [Doctoral dissertation, RMIT University].

Alhazmi, A. H. J., Islam, S. M., & Prokofieva, M. (2025). The impact of artificial intelligence adoption on the quality of financial reports on the Saudi Stock Exchange. International Journal of Financial Studies, 13(1), Article 21. https://doi.org/10.3390/ijfs13010021

Alhazmi, A. H. J., Islam, S. M., & Prokofieva, M. (2026). The CMA Agentic Platform: Autonomous asset verification and algorithmic auditor governance. FinTech, 5(2), Article 55.

Aliraqi, A. M. (2026). AI-driven compliance and risk management in finance: A comparative study of global and Arab regulatory approaches. Journal of Development Studies & Entrepreneurship, 4(2), 1–17.

Almasria, N., Aldboush, H., Al-Hindawi, K., Alomari, A. Y. A., Shiyyab, F. S., & Alharbi, K. M. (2026, April). Ethical oversight and governance in AI-driven digital auditing: Toward sustainable innovation. In 2026 International Conference on Artificial Intelligence for Sustainable Engineering and Innovation (AISEI) (pp. 1626–1632). IEEE.

Alruwaili, T. F., & Mgammal, M. H. (2025). The impact of artificial intelligence on accounting practices: An academic perspective. Humanities and Social Sciences Communications, 12(1), Article 1197. https://doi.org/10.1057/s41599-025-05397-1

Alsayed, H., & Alahmari, H. (2024). Artificial intelligence's impact on the quality of external auditor reports in Saudi domestic and international audit companies. Journal of Economics, Finance and Accounting Studies, 6(5), 98–107. https://doi.org/10.32996/jefas.2024.6.5.10

Asif, D. M., & Khan, D. M. A. (2026). The Impact of AI-Enabled HRM Practices on Employee Work Engagement: The Moderating Role of Employee Digital Literacy. Journal of Business Insight and Innovation, 5(4), 91–103. https://doi.org/10.63544/jbii.v5i4.220

Attia, E. F., Bamhraz, A., Binsaddiq, R., & Qoth, A. S. (2026). Artificial intelligence adoption in accounting information systems: Empirical evidence on performance effects in Saudi Arabia. Journal of Business and Socio-Economic Development, 6(4), 413–429.

Bahlool, R., Hewahi, N., & Elmedany, W. (2026). Performance, fairness, and explainability in AI-based credit scoring: A systematic literature review. Journal of Risk and Financial Management, 19(2), Article 104.

Elfaki, M., Al-Qadi, N. F., Saeed, A., Magboul, I., Amin, H. A., Abdalrahman, N. A., & Algeilani, E. M. (2026). Artificial intelligence adoption in finance: Antecedents, perceived quality outcomes model in a banking sector. Applied Mathematics, 20(1), 297–310.

Falah Alroud, S., Aljabr, M. A., & Al-Shorafa, A. J. (2025). The influence of artificial intelligence on electronic audit evidence: Exploring the mediating role of digital transformation: Evidence from Jordanian export firms. EDPACS, 70(2), 1–40.

Ghaleb, M. M. S., & Pantamee, A. A. (2024). Delving into the corporate governance labyrinth: Exploring the impact of mechanisms on earnings management practices in Saudi Arabian companies. Cuadernos de Economía, 47(133), 1–10.

Ghozi, H. S. A. (2024). Artificial intelligence in internal auditing: Enhancing decision-making and audit quality in the Saudi accounting sector. Decision Making: Applications in Management and Engineering, 7(2), 678–694.

Gulam, G. (2026). Artificial intelligence driven decision support systems for Vision 2030 smart governance in Saudi Arabia.

Helmi, D., & Barhdadi, A. (2025). Impact of artificial intelligence and machine learning adoption on audit quality and risk detection in Moroccan auditing firms. Management Control, Auditing and Finance Review, 2(4), 67–82.

Ibrahim, U. (2025). The impact of artificial intelligence on accounting practices: An academic perspective. Big Data, 2(1), 10–17.

Johri, A., Sayal, A., Chong, K. M., Khoja, M., Jha, J., & Tyagi, N. (2026). Enhancing audit quality and reducing costs: The impact of AI in banking and financial services. Frontiers in Artificial Intelligence, 8, Article 1718854.

Khan, F., Ullah Jan, S., & Zia-ul-haq, H. M. (2025). Artificial intelligence adoption, audit quality and integrated financial reporting in GCC markets. Asian Review of Accounting, 33(3), 464–495.

Mahmood, T., & Asif, M. (2024). Prediction of energy efficiency for residential buildings using supervised machine learning algorithms. Energies, 17(19), Article 4965. https://doi.org/10.3390/en17194965

Mannan, A., Malik, N. A., Noor, M. H., Ali, M., & Siddique, M. (2026). Dynamic approaches to machine learning ethics in software engineering: A comparative exploration of Pakistan and Saudi Arabia. International Journal of Advanced Research, 3(2), 307–328.

Mark, M., Sultan, K., Xiong, G., & Ren, Y. (n.d.). Evaluating fairness and algorithmic bias in Saudi automatic license plate recognition datasets.

Mohammed, A. F. A., & Rahman, H. M. A. A. (2024). The role of artificial intelligence (AI) on the fraud detection in the private sector in Saudi Arabia. Journal of Arts, Literature, Humanities and Social Sciences, (100), 472–506.

Mukarker, E. (2025). Examining the role of artificial intelligence in auditing and fraud detection: The moderating effect of machine learning within accounting information systems. EDPACS, 1–13.

Nokhal, A. M., Sakran, N. E., Atia, F. A., Roushdy, N., Muhammad, M. M., & Aqili, K. I. (2026). Machine learning and external auditors' predictive capacity in detecting financial statement fraud: A hybrid governance and empirical framework.

Omar, M. A., Gomaa, I. I., Sabry, S. H., & Moubarak, H. (2026). Artificial intelligence's role in predicting corporate financial performance: Evidence from the MENA region. Journal of Risk and Financial Management, 19(1), Article 51.

Owies, K. A. (2025). The role of artificial intelligence in developing accounting: Automating processes and enhancing financial reporting in Saudi Arabia. Management, 29(1), 239–255.

Rawashdeh, A. (2024). A deep learning-based SEM-ANN analysis of the impact of AI-based audit services on client trust. Journal of Applied Accounting Research, 25(3), 594–622. https://doi.org/10.1108/JAAR-04-2023-0113

Shaikh, F. A., & Asif, M. (2026). Human-AI collaboration in accounting: A quantitative analysis of generative AI's effects on professional judgement, decision-making, and financial reporting quality. Responsible Education, Learning and Teaching in Emerging Economies, 8(2), 36-54. https://doi.org/10.26710/relate.v8i2.3955