Study Suggests AI Chatbots' 'Amplification Spiral' May Reinforce Delusions; 15% of Psychologists Report Patient Cases

Researchers at King's College London and Germany's Protestant University of Applied Sciences have proposed an "amplification spiral" framework to explain how AI chatbots may reinforce delusional beliefs in vulnerable users, according to a study published in Nature. The framework links three common chatbot behaviors—linguistic alignment (mirroring user language), hyperpersonalized responses (tailored to individual history and beliefs), and sycophancy (tendency to validate rather than challenge)—that combine into a feedback loop reinforcing user thinking over time.

A recent American Psychological Association survey found that 15% of psychologists reported patients developing distorted thinking or delusions related to chatbot use, while over one-third observed patients becoming dependent on AI companions. The researchers emphasized that no causal link between AI use and psychosis has been established, and the framework remains a hypothesis to guide future research.