The words we use and our writing styles can reveal information about our preferences, thoughts, emotions and behaviors. Using this information, a new study has developed machine learning models that can detect antisocial behaviors, such as hate speech and indications of violence, from texts.