ABSTRACT:The rapid expansion of social media platforms such as Facebook, Instagram, X, and WhatsApp has transformed digital communication, online branding, and commercial engagement. However, this rapid digital connectivity has simultaneously created a fertile environment for phishing attacks, malicious shortened URLs, fake login portals, and credential harvesting campaigns. Phishing attacks on social platforms are particularly dangerous because attackers exploit trust relationships, viral content, urgency, emotional triggers, and brand impersonation to deceive users. Traditional blacklistbased detection methods often fail because phishing URLs mutate rapidly, use dynamic redirects, and exploit newly registered domains. Machine Learning (ML) offers a scalable defense mechanism by learning URL patterns, lexical structures, host-based features, and behavioral signals that distinguish malicious links from legitimate content. This paper presents a comparative analysis of major machine learning algorithms used for phishing detection, including Logistic Regression, Decision Trees, Random Forest, Support Vector Machines, Naive Bayes, and Gradient Boosting models. The research evaluates algorithm performance across phishing datasets commonly available on Kaggle and discusses detection accuracy, false positive rate, computational complexity, and deployment suitability for enterprise-scale social media cybersecurity systems. The findings suggest that ensemble learning approaches outperform basic classifiers due to their ability to capture non-linear fraud indicators while maintaining stability under evolving attack conditions.

KEYWORDS: Cybersecurity, Phishing Detection, Machine Learning, Social Media Security, URL Classification, Random Forest, Support Vector Machine, Digital Identity Protection, Social Engineering, Feature Engineering.

CYBERSECURITY IN THE AGE OF SOCIAL MEDIA: COMPARATIVE ANALYSIS OF MACHINE LEARNING ALGORITHMS FOR DETECTING PHISHING LINKS ON SOCIAL PLATFORMS

ISHITTA BHAVSAR, BHARATI GUPTA