Can AI Interactions Contribute to Psychotic Symptoms?

The Guardian’s Science Weekly podcast examined “AI psychosis,” a term used to describe delusions emerging in people who excessively use chatbots. Host Madeleine Finlay spoke with Dr. Hamilton Morrin, psychiatrist and researcher at King’s College London, whose paper explores how AI interactions may contribute to psychotic symptoms. Unlike schizophrenia, reported cases often involve delusions rather than full psychosis, typically falling into three themes: hidden truths about reality, interacting with god-like AI beings, or intense emotional attachment to chatbots. Reinforcement learning that rewards agreeable responses can amplify these beliefs, co-creating delusions. While evidence doesn’t suggest AI is causing widespread new cases, risks exist for vulnerable individuals. Morrin emphasizes the importance of safeguards, clinician involvement, and awareness to prevent emotional overdependence and unsafe AI responses.

The Guardian, Madeleine Finlay and Tom Woolfenden – ‘AI psychosis’: could chatbots fuel delusional thinking? – podcast – 28th August 2025