Bridging the AI Policy Gap in Medical Education: Assessing the Lack of Standardised Guidelines in US Medical Schools

PURPOSE: The purpose of this study is to analyse the AI policies enacted by US medical schools for medical students. METHODS: A cross-sectional document analysis was conducted in July 2024 of publicly available AI-related policies for all 199 accredited US medical schools and their affiliated larger university systems (LUS). Using institutional search tools, we identified student handbooks and policy webpages. Documents that set behavioural expectations for student AI use were coded by two reviewers for policy source, student specificity, content scope and enforcement. Descriptive statistics and chi-square or Fisher's exact tests compared policy presence by programme type and geographic region. RESULTS: Of 199 US medical schools, 121 (61%) had an AI policy. Thirty-one policies originated from medical schools and 90 from a LUS with some overlap. Medical school level policies were more often targeted to medical students and addressed AI use in coursework, assessments, research and clinical documentation, emphasising academic integrity, ethical use and data privacy. LUS policies were typically generic, applying to all university students. Allopathic (MD) schools (83%) and schools in the South (34%) were more likely to have AI policies (p < 0.001). CONCLUSIONS: Many schools rely on generic LUS policies rather than medical student-specific AI guidance. This gap may leave learners without clear expectations for appropriate AI use in coursework and clinical training. Our descriptive findings offer a baseline foundation for strengthening AI governance. Tailored, student-specific and regularly updated policies are essential to support responsible AI use while preserving academic and professional standards.