Purpose : With task-specific knowledge taken into account as a moderating factor, this study attempts to investigate the impact of big data, computer-asisted audit techniques (CAATs), and auditor religiosity on fraud detection. The growing need for insight into the behavioral and technological components that support efficient fraud detection in audit procedures served as the basis for the study. Method : Using an associative quantitative approach, the study was carried out at the State Development Audit Agency’s Representative Offices in Sumatra. A straightforward random sample method was used to gather data, and 220 questionnaires were filled out. Structural Equation Modeling (SEM) was used in the investigation to determine the correlations between the variables. Findings : According to the t-test results, auditor religiosity significantly improves fraud detection. Big data and CAATs, however, did not demonstrate a statistically significant impact. Moreover, task-specific information serves as a predictive modifiers rather than a moderating variable as first proposed. With a termination value of 0.974, the model had a very small effect size. Novelty : The study is innovative since it examines behavioral, technical, and cognitive aspects of the auditing profession in a comprehensive manner. Additionally, it offers fresh perspective by reframing task-specific knowledge as a predicting component rather than a moderating variable, with important ramifications for enhancing fraud detection techniques.