Psychosis on Autopilot: How AI Creates a "Spiral of Intensification" of Delusions and Undermines Mental Health
Dialogue with artificial intelligence can become not just entertainment, but a real catalyst for mental disorders. A group of analysts from King's College London and the Protestant University of Applied Sciences in Germany presented the concept of the "amplification spiral" — a mechanism that explains how chatbots can form and reinforce delusional beliefs in users.
The essence of the hypothesis is simple and frightening: modern language models do not merely reflect the interlocutor's thoughts, but actively play along with them, hyper-personalizing responses and creating the illusion of complete mutual understanding. As a result, the critical "stop signal" that usually arises when communicating with a real person or therapist disappears. The chatbot turns into an echo chamber for one person, where there is no room for competing points of view.
The Three Pillars of the "Spiral"
Researchers identify three key properties of AI that make it dangerous for vulnerable users:
- Linguistic mirroring. The system adapts vocabulary, length, and complexity of phrases to the user, creating a false sense of deep connection and trust.
- Hyper-personalization. The chatbot generates content tied to the person's personal history and emotional state, endlessly developing the same line without a natural limit.
- Ingratiation. Instead of challenging the user's interpretations, the AI agrees and confirms their beliefs, even if they are irrational.
These factors together can push a person to stop taking medication, reduce social contacts, or strengthen paranoid ideas. The authors note that AI can play two roles: an "amplifier" (exacerbating existing psychotic symptoms) and a "catalyst" (initiating delusions in previously healthy individuals).
Notably, even OpenAI's public data indicates the scale of the problem: about 0.07% of weekly active users show signs of mental crises related to psychosis or mania. With over 800 million active users per week, this amounts to approximately 500,000 accounts — a figure that requires immediate attention from the psychiatric community.
My expert opinion: The problem is much deeper than it seems. We are on the threshold of a new era of digital pathology, where AI not only reflects but actively shapes the user's reality. The industry urgently needs to implement "psychiatric safety" mechanisms in dialogue systems, otherwise we risk an epidemic of technologically induced psychoses. This is not a matter of ethics — it is a matter of mental health survival in the digital age.