People are increasingly using ChatGPT and similar AI chatbots as mediators for relationship conflicts and mental health support [1].
This trend reflects a growing reliance on low-cost, accessible technology to address loneliness and interpersonal strife. However, the shift raises significant concerns regarding the safety of mental health care and the ability of software to handle complex human emotions.
Research from Brown University indicates that these AI systems routinely break core ethical standards of mental health practice [4]. Experts said chatbots lack the ability to read body language and understand cultural context, both of which are critical for effective therapy [3]. Because AI cannot perceive power dynamics or non-verbal cues, it may overlook dangerous or nuanced situations within a relationship [3].
Despite these warnings, some users report that AI mediation helps them feel less divided during arguments [1]. The appeal often lies in the constant availability of the tools. In the UK, reports indicate that one in three adults [5] and one in four teens [5] have turned to AI chatbots for mental health support.
Critics said that the perceived success of AI mediation is superficial. One couple participant described a situation where the AI conveyed a sentiment that the partner did not actually hold, illustrating a gap in true understanding [2].
While some professionals suggest AI can provide basic emotional support in specific contexts [3], they said that it cannot replace a licensed therapist. The lack of professional oversight means there are no safeguards to protect users from harmful or incorrect advice [4].
“"These systems routinely break core ethical standards of mental health practice."”
The rise of AI-driven mediation suggests a widening gap between the demand for mental health services and the availability of affordable professional care. While AI offers a scalable, immediate tool for conflict resolution, its inability to interpret human nuance creates a risk where users may mistake algorithmic pattern-matching for genuine emotional intelligence or clinical validity.
