Due to the widespread use of social media, scammers attempt to deploy malicious social bots that produce fake tweets, attempt to build relationships with other users by pretending to be followers or attempt to create several fake accounts that engage in malicious actions. Additionally, they frequently post malicious URLs that direct real people to malicious web servers. Therefore, it is crucial to distinguish between legitimate accounts and bot accounts. It has been found that profile-based features and URL features, such as redirected URLs, spam data, frequency of URL sharing, etc., are better indicators of bots than social factors. In this study, we propose a novel method that exposes malicious bots on social networks by utilizing profile-based attributes and Deep Learning algorithms. We apply the aforementioned model to the Twitter data set and see that it performs better than other methods. We also made an effort to create a web application that might demonstrate that the aforementioned strategy performs better than other models that are already in use.