Fake Social Media Profile Detection using Machine Learning Algorithms

social media has reshaped global interactions, offering unprecedented networking opportunities for individuals and businesses. However, its widespread reach also facilitates the rapid spread of harmful content, including hate speech directed at race, gender, religion, and disabilities, potentially causing significant emotional harm. To mitigate these challenges ML and DL techniques are becoming essential tools in identifying fraudulent social media profiles. These advanced methods analyse behavioural patterns, account details, and interactions to detect anomalies that indicate deceptive activity. ML algorithms classify suspicious accounts based on predefined features, while deep learning models—such as neural networks—process vast amounts of data to uncover more complex fraudulent tactics. As fraudsters evolve their strategies, AI-driven solutions continue to improve, enhancing social media security and protecting users from misinformation and scams. This study evaluates six ML models Neural Networks (NN), Naive Bayes (NB), Logistic Regression (LR), XGBoost (XGB), Random Forest (RF), and Support Vector Machine (SVM)using real-time datasets that have undergone pre-processing to optimize feature extraction. Among the evaluated models, SVM achieved the highest accuracy, surpassing both SVM and NB in precision, recall, and F1-score