Phishing attacks on cyber applications are more popular attacks and severe problems. Now a day mostly 60% of email links with fraud links that navigate users to clone websites. These sites are attacker’s sites where they clone websites to hack user personal details. Phishing is a thread on online applications to hack user login details digitally. Phishing web documents are looks like original website and user are not in a position to identify clone web documents. Cybercriminal change user details with in no time to hack users valuables. All these reasons force the need of security mechanism that will protect users from phishing attack. Many techniques were introduced by various researchers to identify phishing web documents using heuristic technique and black listed technique. In this research, we proposed an efficient protection mechanism that predicts users from phishing attacks by using ML approaches like logistic regression (LR), random forest (RF), support vector machine (SVM), naïve Bayesian classification (NB), and Decision Tree (DT) algorithms. From experiments calculated confusion matrix with that all performancematrices are evaluated. SVM shows 98 percentage of accuracy.