Pediatric GenAI Policy Calls for Validation, Human Oversight, and Bias Safeguards

A policy statement published in Pediatrics outlines recommendations for developing and implementing generative artificial intelligence (GenAI) tools in the care of children and adolescents. Potential applications include clinical decision support, documentation, and medical education. The statement is policy guidance rather than a clinical trial; the available abstract provides no sample sizes or quantitative performance estimates.

The statement notes that real-world validation remains limited and that studies have found large language models often underperform in pediatrics compared with adult medical specialties. It highlights concerns about accuracy, reliability, sustainability, and durability, as well as the potential for biased or nonrepresentative training data to exacerbate health disparities.

Recommendations include prioritizing diverse pediatric datasets, addressing bias, and implementing strong privacy and security safeguards. Health care institutions should establish rigorous validation protocols, comply with privacy regulations, and maintain clear human oversight. Regulatory and oversight bodies should require pediatric-specific evaluations and postmarket surveillance.

For clinicians, the statement emphasizes GenAI literacy, interdisciplinary governance, transparency, and reasonable disclosure of GenAI involvement in patient care. It calls for adaptive policies that support ethical, equitable, evidence-based implementation. Full text was unavailable, limiting assessment of the supporting evidence and detailed recommendations.