Publication
Published on: May 12, 2026
Published in: NEJM AI
Description: Barbara Bierer and Trevor Baker of the MRCT Center, with co-authors Rishabh Goel and Sammy Mustafa of Bond Health and Harvard Medical School’s Department of Biomedical Informatics, published “Large Language Models in Informed Consent — Opportunities, Evidence, and Challenges” in NEJM AI. Informed consent forms remain lengthy, cognitively demanding, and written at a level beyond most participants’ reading level, contributing to gaps in perceived understanding and recall of key study elements. This narrative review maps the emerging landscape of LLM applications for informed consent across seven domains: glossary generation, plain language rewriting, visual aid generation, accessibility and document formatting, translation, teach-back and comprehension assessment, and chatbot-assisted consent. For each domain, the authors synthesize the available evidence, surface principal failure modes such as hallucination, omission, and demographic bias, and outline mitigation strategies and considerations for integration with research ethics committee review.
