AIHealthcare Analytics

National Governance Initiatives

The French Conference of Deans of Medical Faculties (CDD) developed and unanimously adopted a national charter for responsible generative AI use in health education (November 26, 2025). The charter is structured around six components: general principles (complementary use, transparency, traceability, human accountability); authorized uses (text structuring, synthesis, linguistic editing, translation, supervised code generation); prohibited uses (data fabrication, plagiarism, substitution for critical reasoning, entry of personal data into non-sovereign systems); defined responsibilities; oversight mechanisms with proportionate sanctions; and forward-looking capacity-building measures.

U.S. Institutional Policy Landscape

A mixed-methods study of ASPPH member schools found only 18 of 155 schools had a generative AI use policy. Policies predominantly focused on academic integrity at the intersection of student and faculty AI use. Data privacy, security, research ethics, and pedagogical guidance were underrepresented. The evidence base for AI governance in public health education is described as in its "relative infancy."

Clinical Placement and Student Use

A cross-sectional French university survey (n=388) found 52.6% of health professions students used GenAI during clinical placements. Among users, 23.5% reported at least one disclosure of patient-identifying information, and 47.1% reported processing real medical information with GenAI. Information retrieval (77.9%), bibliographic search (74.5%), and translation/rephrasing (71.1%) were the most common uses. Midwifery students showed lower uptake (26%) versus other disciplines.

Key Risks and Recommendations

Citation fabrication — the generation of plausible but nonexistent references — is identified as a distinct academic integrity challenge differing from plagiarism in that it is often unintentional, nearly undetectable by standard tools, and reveals gaps in evidence-based scholarship competency. AI detectors falsely label non-native English speaker (NNES) writing as AI-generated in 50.2%–61.3% of cases, creating equity risks in assessment. Institutional responses should include AI literacy frameworks, revised academic integrity policies, and embedded source verification curricula.