Citation quality varied sharply across 6 chatbots; keyword relevance hallucinations reached 61.6%, underscoring the need to verify AI-generated references. #MedicalChatbots #ReferenceAccuracy #Hallucination doi.org/10.2196/54345