In the 20th century, most people are depends on the internet. So the chances of cyber fraud are increased rapidly. The purpose of these frauds is to get the information used by individuals and organizations for transactions and misuse of information. We proposed social media application for detecting phishing URLs or IP addresses based on Extreme Learning Machine Algorithm. Where people can communicate with each other to get more information about these frauds. We proposed an application based on machine learning techniques to detect phishing websites. It compares with 23 parameters and gives us one output based on these parameters. The output is in the form of normal, suspicious, and phishing.