The rapid advancements in mobile phone systems and programs that provide free instant messaging (IM), short message service (SMS), and the accommodation of conveying millions of messages with virtually no delay and zero cost through Wi-Fi or 3G(third generation) has led to the increasing popularity of IM and SMS.The requisite for these advancements is an automatic classification system for expeditious relegation of the received messages to detect the suspicious message. This work proposes the use of a detection model in which social media messages are classified as predefined classes labeled“suspicious” and “not suspicious.”The proposed system attemptsto solve this problem through three classifiers:Level Based Feature Content(LBFC) classifier, Naive Bayesian(NB) classifier, and IterativeDichotomiser 3(ID3) classifier.This system works offline, after collecting the messages online, saving them and then inputting them into the proposed system.In the LBFC classifier, the content feature is divided into four levels to detect suspicious classes.The second classifier presents an NB classifier and the the third is ID3 classifier is capable of identifying Viber messages as suspicious or non-suspicious, predicated on the content of these messages.From the experimental work, good results are achieved from the first classifer t features (Accuracy=0.882143%) and second testing using term frequency (TF)-based NB classifier (Accuracy=0.942857%), while results are achieved using TF-based ID3 (Accuracy= 0.957143%).