AI-Driven Techniques for Detecting Malicious and Fake Profiles Across Social Media Networks

Abstract

With billions of users worldwide, social media platforms have become essential channels for communication and information sharing. However, the rapid expansion of these platforms has also led to a significant increase in fake accounts used for malicious activities such as phishing, spam campaigns, fraudulent advertising, and the spread of misinformation. Traditional detection approaches based on rule-based filtering and manual moderation are insufficient to handle large-scale data and are ineffective against sophisticated fraudulent behaviors. To address these limitations, this study proposes an artificial intelligence-driven fake account detection system that integrates machine learning techniques with behavioral feature analysis. The system evaluates user profile attributes such as follower count, account age, posting frequency, and username characteristics to classify accounts as genuine or fake. Multiple machine learning models, including Random Forest, Support Vector Machine, and Neural Networks, are trained and evaluated for performance comparison. A Flask-based web interface enables real-time prediction and visualization of results. Feature scaling, encoding, and automated model selection further enhance detection accuracy and system efficiency. Experimental results demonstrate improved classification performance compared to conventional approaches, supporting scalable deployment in real-world environments. The proposed framework provides an intelligent, automated, and transparent solution for strengthening security in online social networks.