Malicious Uniform Resource Locator (URL) is a common and serious threat to cyber security. Malicious URLs host unsolicited contents (spam, phishing, drive-by exploits, etc.) and lure unsuspecting internet users to become victims of scams such as monetary loss, theft, loss of information privacy and unexpected malware installation. This phenomenon has resulted in the increase of cybercrime on social media via transfer of malicious URLs. This situation prompted an efficient and reliable classification of a web-page based on the information contained in the URL to have a clear understanding of the nature and status of the site to be accessed. It is imperative to detect and act on URLs shared on social media platform in a timely manner. Though researchers have carried out similar researches in the past, there are however conflicting results regarding the conclusions drawn at the end of their experimentations. Against this backdrop, four machine learning algorithms:Naïve Bayes Algorithm, K-means Algorithm, Decision Tree Algorithm and Logistic Regression Algorithm were selected for classification of fake and vulnerable URLs. The implementation of algorithms was implemented with Java programming language. Through statistical analysis and comparison made on the four algorithms, Naïve Bayes algorithm is the most efficient and effective based on the metrics used.