Study reveals AI mental health chatbots breach ethical codes

JAKARTA (ANTARA) - A study from Brown University has revealed AI chatbots used for mental health therapy may breach professional psychology ethical codes. According to Psychology Today on Thursday, the research found large language models (LLMs) often fail to adhere to mental health ethical standards such as those set by the American Psychological Association (APA). Lead author Zainab Iftikhar stated psychotherapy cannot be treated as a simple computational task, requiring strict adherence to ethical standards and professional conduct codes. The 18-month study involved seven trained peer counsellors and three licensed clinical psychologists evaluating AI chatbot behaviour in counselling sessions. From 137 sessions analysed, researchers identified 15 ethical violations grouped into five categories: inadequate context understanding, poor therapeutic collaboration, misleading empathy, unfair discrimination, and weak crisis and security handling. Researchers noted AI chatbots often oversimplify users’ life experiences, dominate therapy conversations, and display manipulative empathy. AI also struggles to handle sensitive issues such as trauma, violence, and suicidal ideation. The study emerges as AI use in mental health services grows globally. Another study published in AI & Society journal in February 2026 shows increasing public openness to using AI as mental health advisors. The AI mental health industry is expected to expand rapidly, with research firm Grand View Research estimating the market to reach $9.12 billion by 2033. While offering accessible real-time support, researchers warn that using AI as therapists poses serious risks without proper regulation. They urge clearer legal guidelines to minimise potential dangers for users of AI-based therapy services.