LLM Provider Terms Block Core Security and Social Science Research Methods

New analysis of major LLM providers' Terms of Service reveals substantial variation in usage restrictions that unintentionally prevents legitimate academic research. Security researchers cannot conduct red-teaming; computational social scientists cannot systematically test bias or content moderation. Broad prohibitions meant to prevent misuse now block essential methodologies across psychology, social science, and security disciplines, fragmenting research across platforms with inconsistent policies.