Study links AI chatbots to increased delusional thinking

A new study published in The Lancet Psychiatry suggests a possible link between AI chatbot use and increased delusional thinking in some users. According to the researchers, people already predisposed to psychotic disorders may be particularly vulnerable. The authors call for comprehensive clinical studies of such systems involving mental health professionals.

The review's author, psychiatrist and researcher at King's College London Hamilton Morin, analyzed about twenty media reports describing cases of so-called "AI psychosis." The study aimed to determine whether chatbots might amplify or trigger delusional beliefs. Morin noted that some systems tend to agree with users' unusual or exaggerated claims, sometimes even expanding on them in conversation.

Researchers point out that delusional ideas can take various forms—from beliefs in one's own exceptionalism to persecution complexes or romantic delusions. Chatbots most often reinforce "grandiose" beliefs, since their algorithms focus on maintaining dialogue and frequently avoid contradicting the user. In several described cases, systems used cryptic phrasing and hinted at users having special "spiritual roles" or connections with cosmic entities.

However, scientists emphasize that there is no evidence that artificial intelligence can cause psychosis in people without predisposition. Rather, interactive systems may accelerate the development of existing doubts or suspicions. Previously, people sought confirmation for unusual ideas in books or videos, but now a chatbot can support such reasoning almost continuously, potentially amplifying the effect.