Self-Reported Side Effects of Semaglutide and Tirzepatide in Online Communities

This collection highlights several emerging risks at the intersection of social media, generative AI, and information operations. Key findings: (1) LLMs can reliably infer users’ political alignment from non-explicit linguistic cues in online conversations—creating a high-risk privacy and profiling vector for targeted persuasion, surveillance, or repression; (2) Character.AI hosts millions of user-created public chatbots, raising large-scale moderation, youth-safety, and data-leakage concerns where user-generated bot content can propagate harmful tropes or be repurposed for influence; (3) a 6‑