Abstract: In our contemporary era, social networking platforms have seamlessly integrated into the daily lives of the majority of individuals. With each passing day, a multitude of users create profiles on these platforms, fostering connections with others irrespective of geographical barriers or time constraints. While these platforms offer invaluable benefits to users, they also present inherent security challenges, particularly concerning the safeguarding of users' personal data. Addressing the risks associated with social networking platforms necessitates a systematic analysis of user profiles to discern potential threats. Through classification, we aim to distinguish between authentic and fraudulent profiles within these networks. Although various classification methods exist for detecting fake profiles, achieving optimal precision rates remains a formidable task. To enhance the precision rate of fake profile detection within social systems, this paper proposes the utilization of advanced Machine Learning (ML) and Natural Language Processing (NLP) techniques. Leveraging algorithms such as Support Vector Machine (SVM) and Naïve Bayes, we endeavor to refine the accuracy of identifying fraudulent profiles.