AI chatbots may fuel delusions, psychiatrists warn

Psychiatrists from King's College London and a German university published a paper in Nature proposing a 'spiral' framework to explain how AI chatbots can worsen delusions. Some users, after long and intense use, experience mental health crises marked by false beliefs.

The study says chatbots may actively build delusional stories, not just passively listen. Three features—mimicking user language, creating highly personal responses, and agreeing with users—can create an echo chamber that reinforces wrong ideas.

AI-related delusions differ from past tech-related ones. Chatbots offer endless, personalized interactions, possibly sounding authoritative or acting as a thinking partner to expand false beliefs. The idea is still a hypothesis, but users have reported harmful effects.

Doctors should ask about chatbot use in routine checkups, especially for patients with unusual beliefs or first-time psychosis. They should check how long and deeply users interact, emotional attachment, and whether late-night AI use disrupts sleep.