Building Social World Models with Large Language Models

This feed contains research (June 2026) across social modeling, content-moderation, and ML methods with several security-relevant themes. Key items: (1) Social World Models (SWM) use LLMs and prediction-market data to learn state-transition functions for social beliefs — enabling accurate prediction of belief dynamics (dual-use risk: targeted influence, social-manipulation campaigns). (2) “Great Disappearance Acts” analyzes generative search (RAG) and shadow-banning, highlighting risks to content availability, copyright/competition exposure, algorithmic suppression, and regulatory responses (A