The Dark Side of Anonymity: Measuring Online Hate on 4chan

This research examines the prevalence and impact of hate speech on 4chan’s politically incorrect board (/pol/), a largely unmoderated online space. Using advanced Natural Language Processing (NLP) techniques, including RoBERTa and Detoxify, the study analysed a dataset of 500,000 posts to quantify the types and frequency of hate speech. The results showed that 11.2% of posts contained hate speech, with racism, religious hatred and sexism being the most common forms. This work contributes to our understanding of how online hate spreads in less regulated spaces, and highlights the need for AI-driven moderation tools and digital literacy initiatives. Its findings are vital for policymakers and platform developers in developing regulations such as the UK’s Online Safety Act, and provide a basis for better content moderation and safety standards in digital environments.

More details in the related outputs

For the details of the paper, read our blogpost here.