This study develops and validates a cognitive auditing model based on generative artificial intelligence for detecting hidden patterns of financial fraud in intelligent organizations. Given the increasing complexity of financial structures and the rise of multi-layered and data-driven fraud schemes, traditional auditing methods are no longer sufficient for identifying anomalous behaviors. This research adopts a mixed-methods (qualitative–quantitative) approach. In the qualitative phase, thematic analysis and interviews with 18 experts were conducted to extract the main dimensions of the model, including data-driven infrastructure, cognitive capabilities, generative AI, information processing, anomaly detection, and intelligent decision-making. In the quantitative phase, data were collected from 320 valid questionnaires and analyzed using Partial Least Squares Structural Equation Modeling (PLS-SEM). The results indicated that the proposed model has an acceptable fit and all relationships among constructs are statistically significant. Findings show that data-driven infrastructure, through enhancing cognitive and generative capabilities, plays a crucial role in improving fraud detection accuracy, reducing financial risk, and increasing financial transparency. Moreover, generative AI, with its ability to analyze both structured and unstructured data, enables the identification of hidden patterns and the prediction of fraudulent behaviors. Overall, the results suggest that cognitive auditing can serve as an innovative and effective approach for strengthening auditing systems and combating financial fraud in intelligent organizations.