Twitter is known as a platform with great influence on public opinion. Many companies operate fake twitter users (or “handles”) with the sole purpose of influencing public opinion. Such handles are tagged as “Trolls”. In order for Twitter to not block the troll handles, the operating companies of such handles try very hard to disguise them as regular handles. In this project we had built a machine learning system which classifies twitter handles as Trolls based on their tweets. In order to train our learning models we used a dataset from Kaggle which contains data on nearly 3 million tweets. To find the best results for the task we tested 6 different algorithms (Logistic regressions, KNN, Classical Decision tree, XGBoost, Neural Network, Bert), and for each algorithm we tried different types of numeric representation of the input data like Bag Of Words, TF-IDF and Word2Vec. Finally, we compared the results of the different algorithms to see which one will be the best for solving the problem. Bert algorithm reached the best results, but in terms of training time vs accuracy Logistic Regression was the most successful for this task.