When Jokes Cross the Line: Analyzing Regular Humor and Dark Humor in YouTube Shorts
For researchers and platform moderators, this work provides a benchmark and empirical analysis of the gray area between humor and harm in short-form video, highlighting the need for context-aware moderation.
The paper introduces TwistedHumor, a dataset of 1,211 YouTube Shorts with annotations for humor and harm, and finds that dark humor clusters around themes like critique and coping, while regular humor receives more positive audience sentiment and dark humor more mixed and toxic reactions. LLMs perform better on stand-up comedy than shorter jokes.
Video platforms such as YouTube have reshaped how users engage with entertainment and information, emphasizing brief, highly engaging content such as Shorts. Within this ecosystem, certain content occupies a gray area where it remains allowed but may still have unintended negative effects on some audiences. To study this problem, we introduce TwistedHumor, a dataset of 1,211 YouTube Shorts paired with 33,041 related comments, with hand annotations for humor presence, humor type, harm, topic, rhetorical devices, and stand up context. Beyond dataset creation, we present a multi view analysis of how humor and harm appear in short form social media. Using LLooM based concept induction over video descriptions, we find that dark humor frequently clusters around themes of critique, coping, awkwardness, and identity expression rather than appearing as a single uniform category. We further analyze audience response through linked comments and show that regular humor is associated with more positive sentiment, while dark humor receives more mixed, neutral, and sometimes more toxic reactions. Finally, we evaluate large language models against human annotations and find that they perform better on stand up comedy compared to shorter jokes. Together, these results position TwistedHumor not only as a new benchmark, but as an empirical study of the gray area between humor and harm in short form video, highlighting the need for context aware moderation and more robust multimodal evaluation.