1 min
New AI model predicts harmful videos before they go viral
Certain short-form videos on major social media platforms can trigger suicidal thoughts among vulnerable viewers, according to new research led by the University of Delaware's Jiaheng Xie. His team developed an AI model that can predict and flag these videos. In the study, published Nov. 17 in Information Systems Research, UD professor Jiaheng Xie and his co-authors showed how AI can assist with safety by predicting high-risk videos before they go viral by looking through both the content and what people write in the comments. • Distinguish what creators choose to post from what viewers think or feel after watching. • Separate known medical risk factors from emerging social media trends, such as viral heartbreak clips or challenges that may influence teens. • This system is especially novel because it predicts the high-risk videos before they reach large audiences. Xie is available for a Zoom interview to share how the model was developed and how it could potentially change the way platforms such as TikTok moderate.



