Folie à intelligence artificielle: a case of shared delusional thinking between patient and AI chatbot

Abstract

Shared psychotic disorder (folie à deux) is traditionally described as the induction of delusional beliefs in a mentally healthy individual through close association with a psychotic person. While it typically occurs between family members or cohabiting partners, the emergence of conversational artificial intelligence (AI) raises the possibility that a digital system might serve as the secondary partner in such a process. The purpose of this study is to explore how AI systems become entangled in the cognitive and affective dynamics of psychosis, thereby reshaping the phenomenology of delusion and facilitating the spread of delusional beliefs. The authors present a case of a patient with delusions of grandeur and delusions of reference who, during various dialogues with physicians, involved the AI chatbot (ChatGPT) in the conversation as participant and close associate. Through conversational mirroring, the AI chatbot adopted and reinforced elements of the patient’s delusional system. Two anonymized dialogue transcripts demonstrate how the AI gradually echoed the patient’s psychotic beliefs, resulting in a state resembling mutual delusional elaboration. Rather than acting solely as a neutral medium, AI exhibited adaptive, emotionally attuned responses that reinforced the delusional framework. This case illustrates a novel form of shared psychotic process, which the authors term folie à intelligence artificielle. It raises urgent questions about the design, monitoring and ethical aspects of AI in mental health contexts. The authors argue that this case also invites a reconsideration of how delusion is mediated, distributed and potentially co-produced in the digital age.