Energy fraud and electricity theft are serious problems that impact power systems' dependability and result in large financial losses for power distribution corporations. Because smart meters provide a lot of data about electricity consumption, it is challenging to manually detect such fraudulent operations. Using data on electricity consumption, this study suggests a machine learning-based method for identifying power theft. The technology looks for unusual behavior that can point to fraudulent activities by analyzing customer usage trends. Based on their patterns of electricity usage, a number of machine learning techniques, such as Logistic Regression, Decision Tree, Random Forest, and Support Vector Machine (SVM), are used to categorize customers as normal or fraudulent. To increase prediction accuracy and dependability, the dataset is preprocessed and examined. Python and the Django framework are used to create the system as a web-based application that enables users to upload consumption data and obtain fraud detection findings. Machine learning approaches can successfully identify suspicious consumption behavior and increase the accuracy of fraud detection, according to experimental investigation. Power utility firms can monitor electricity use, lower revenue losses, and increase the effectiveness of energy management systems with the help of the suggested solution.