It was only lately realised that the safeguarding of our information is of paramount importance. As part of this work, we examined the use of machine learning, more specifically how machine learning can be used for the detection of security threats using the CICIDS2017 dataset. To clean the data, the missing values were seen to be fixed, then classes were balanced using SMOTE, and lastly the models were optimized by choosing the most important features. Four different machine learning algorithms were applied: SVM which is for Support Vector Machines, Random Forest, Decision Tree, and Naive Bayes. Their performance was then evaluated with help of accuracy, precision, recall value and F1 score. Random forest algorithm turned out to be the most effective showing 94.3% accuracy, while the SVM model achieved 92.5% accuracy. Such outcomes bring out that machine learning can be actually very helpful in making us enhance cybersecurity, and set up much better preventions.