LLM Providers' Terms Are Blocking Academic Research Into Safety and Bias

Major AI companies' service agreements now restrict legitimate academic research. Security researchers cannot test for vulnerabilities; social scientists cannot systematically study bias or content moderation. Broad rules designed to prevent misuse unintentionally block core research methods across psychology, security, and social science, forcing researchers toward inconsistent platforms with varying policies.