AI chatbots are increasingly integrated into mental health care, expanding access to support yet bringing significant ethical, clinical, and design challenges. This integrative review synthesizes empirical studies, reviews, case reports, and media articles from 2015 to September 2025, critically examining the use of both rule-based and large language model (LLM) chatbots. While AI systems show promise for screening, prevention, assessment, treatment, and emotional support, concerns remain about low user retention, privacy risks, algorithmic bias, and the provision of trauma-informed, culturally sensitive care. Phenomena such as “AI psychosis” and emotional dependency further highlight the need for robust risk assessment and regulation. The review underscores the urgency of implementing the Safe Integration of LLMs in Mental Health Care Framework as well as involving vulnerable groups in the co-design process of emotionally intelligent AI chatbots to improve accessibility, safety, and effectiveness.
Luke Balcombe (2025) studied this question.