When Speech Turns Violent Exploring Threat and Violence in Incel Environments

Sammanfattning

This study examines violent speech within blackpill.club, an online community aligned with the Blackpill ideology—a radicalized extension of the broader incel (involuntary celibate) movement. The Blackpill worldview is characterized by deep feelings of rejection, frustration, and anger toward women and society, often resulting in a sense of hopelessness regarding intimate relationships. Within this context, online violence—defined as language that threatens or promotes physical harm toward individuals or groups—represents a growing concern. The study investigates the frequency, targets, and severity of violent speech on blackpill.club, and compares the performance of two AI-based annotation tools, ChatGPT and Grok, against human annotators in detecting and classifying such language. Employing a mixed-method approach and a documentary research design, 1,182 posts from the forum were analyzed to identify patterns of hostility and threat intensity. Findings show that approximately 72% of comments contained violent or threatening expressions, with women as the predominant targets, followed by politicians, LGBTQ+ individuals, and other minority groups. Severity analysis revealed that 36% of the violent posts were high severity (explicit threats or encouragement of serious harm), 29% moderate (indirect threats or harmful suggestions), and 34% low (general hostility or aggression). The results underscore the prevalence of violent discourse within Blackpill spaces and illustrate how such rhetoric perpetuates gendered and societal animosity. While human annotation yielded the most reliable outcomes, ChatGPT demonstrated stronger alignment with human judgments than Grok, though both systems exhibited notable limitations. These findings emphasize the importance of developing improved computational tools and intervention strategies to monitor and mitigate harmful online communication.