AAP Outlines Generative AI Policy for Pediatric Care

The American Academy of Pediatrics (AAP) has issued a policy statement in Pediatrics with recommendations for developing and implementing generative artificial intelligence (GenAI) tools in pediatric clinical care. The AAP Council on Clinical Information Technology and Section on Innovation in Therapeutics and Technology authored the statement.

The authors noted that GenAI tools such as large language models (LLMs) show promise for clinical decision support, documentation, and medical education, but real-world validation remains limited. They reported that LLMs often underperform in pediatrics compared with adult specialties; in one evaluation of board examinations, GPT4 performed significantly worse in pediatrics than in general surgery, internal medicine, and psychiatry. The authors cautioned that biased or nonrepresentative training data could exacerbate health disparities.

The AAP recommends that developers use training datasets that represent the full diversity of pediatric patients, establish validation protocols to detect and mitigate bias across pediatric subpopulations, and protect against re-identification of de-identified data. When information is extrapolated from adult data because no pediatric data were available, the statement says the GenAI output should indicate that. Institutions selecting tools should ensure compliance with the Health Insurance Portability and Accountability Act (HIPAA) and confirm that tools collect only the minimum data necessary.

According to the statement, current GenAI tools require direct human oversight, leaving the pediatrician ultimately responsible for medical decisions. The AAP recommends that regulatory bodies obtain evidence of safety and efficacy in pediatric populations before approving GenAI systems for use with children and adolescents and implement postmarket surveillance that accounts for performance drift.

The statement addresses GenAI tools used by pediatricians to support clinical care, not direct use of AI by children, adolescents, and families. The AAP noted that its guidance does not indicate an exclusive course of treatment or serve as a standard of medical care.

Source: Suresh S, Hunter RB, Periard B, Espinoza J; Council on Clinical Information Technology; Section on Innovation in Therapeutics and Technology. Recommendations for the development and implementation of generative artificial intelligence tools in pediatric clinical care: policy statement. Pediatrics. Published online October 3, 2026. doi:10.1542/peds.2026-079037