فهرست نشریات علمی دسترسی آزاد ایران

This study develops and validates a conceptual model for detecting financial fraud in the financial reporting of firms listed on the Tehran Stock Exchange, emphasizing transparency and financial discipline through artificial intelligence. Based on established theoretical foundations, the model incorporates auditing, corporate governance, managerial, and financial indicators as the principal determinants of fraudulent reporting. Panel data covering the period 2013–2024 were collected and labeled using the adjusted Beneish M-Score (Adj-M-Score). Both conventional statistical methods and machine learning algorithms were applied to assess predictive performance. The results demonstrate that tree-based models, particularly XGBoost, achieve the highest predictive accuracy (AUC ≈ 0.85). Feature importance and SHAP analyses indicate that governance- and behavior-related variables, together with liquidity indicators such as the current ratio and operating cash flow to total assets, are the most influential predictors of fraud. Overall, integrating behavioral, financial, and governance dimensions within an explainable AI framework provides a robust and effective approach for improving financial transparency and detecting fraudulent reporting.