Using a labelled data set comprising of Wikipedia’s Talk page comments, a model was trained to categorize toxic comments into two categories - toxic and threat. This model was then proposed as a means to auto-hide toxic and threat comments from social media platforms (chrome plug-in under development).