Cybersecurity has become a critical issue in the digital era, with evidence in the last 3 years there have been 6 cybercrime cases in Indonesia that attacked servers, one of which was the latest theft of Bank Syariah Indonesia data in May which resulted in the server being paralyzed for 5 days and the impact was that customers could not access the mobile banking application. From the various cybercrime cases that have occurred in Indonesia, we need to know the current trend of public sentiment about it and one of the sources of public sentiment is Twitter. The use of Machine Learning (ML) and Natural Language Processing (NLP) has become a major focus in understanding public sentiment contained in twitter data. This research proposes an approach that combines ML and NLP techniques to detect sentiment in tweets. The method includes a pre-processing stage to clean and transform the tweet text into a vector representation, followed by the application of ML classification model namely Naïve Bayes to identify positive, negative or neutral sentiments from the tweet dataset. This research utilizes a set of collected and annotated tweet data using python to train and test the model. The experimental results show that the proposed approach successfully produces sentiment classification with an accuracy rate of 62%. It can be concluded that the accuracy of the model is still satisfactory with a positive recall value of 74%, meaning that the public sentiment of the tweets still contains words of a positive nature.