This research presents a practice for predicting user originality in password activities using machine learning. The study aims to leverage machine learning to distinguish between authorized and unauthorized users who have obtained the original user’s password and intend to change it. The primary objective is to detect whether the password change task was initiated by the original user or a fraudulent user with knowledge of the password. The results of this work are vary promised as we apply the ‘JRip’ classifier, that results with 175 instances correctly classified and 55 incorrectly classified instances, resulting in a mean absolute error of 0.3081. The accuracy of the ‘JRip’ classifier was indicated by 76.087% for correctly classified instances and 23.913% for incorrectly classified instances. The weighted average metrics for 230 users include a true positive rate of 0.761, a false positive rate of 0.298, a precision of 0.766, a recall of 0.761, an F-measure of 0.763, a (ROC) curve area of 0.740, and a (PRC) curve area of 0.736.