Fake and automated bot accounts have proliferated as a result of social media platforms' explosive expansion, endangering user confidence, information integrity, and platform security. Early detection of such accounts is essential to stopping the spread of malicious activity, spam, and false information. In this research, a method that combines behavioral analysis and graph-based machine learning techniques for the early detection of phony and bot accounts is presented. To detect unusual user activity, the suggested method makes use of characteristics including follower-following patterns, posting frequency, account age, and profile completeness. Furthermore, graph-based user relationships are taken into account in order to capture connectivity patterns that are frequently linked to coordinated bot networks. Kaggle datasets are used in the model's development to ensure diverse and realistic data representations. Accounts are categorized as human, suspicious, or bots using a risk assessment system. The system is built as a web-based application that offers findings that are easy to understand in addition to real-time predictions. The method's efficacy in spotting suspect accounts early on is demonstrated by experimental investigation. For increased accuracy and practical implementation, the suggested method provides a scalable and effective framework that can be expanded with sophisticated machine learning models