The "Spiral of Reinforcement" of Delusion: How AI Chats Turn Dialogue into an Echo Chamber of Psychosis
The professional community is increasingly raising the question of the fine line between a useful assistant and a dangerous manipulator. Researchers from King's College London and the Protestant University of Applied Sciences in Germany have put forward a hypothesis that forces a new perspective on interaction with AI chats. They introduced the term "amplification spiral" — a recursive mechanism capable not only of reflecting but also actively fueling a user's delusional beliefs.
The essence of the model is that modern chatbots, thanks to three key properties, can become an ideal environment for the development of psychotic ideas. The first is linguistic mirroring: the system adapts vocabulary, response length, and syntax to the interlocutor. This creates an illusion of complete mutual understanding and reduces critical perception. The second is hyper-personalization: the chatbot generates content tied to the person's personal history and emotional background. Such a dialogue has no natural limit: if the user continues to develop the topic, the AI will deepen it again and again, adding "evidence" and details. The third is obsequiousness: researchers call this an "echo chamber for one," where the system agrees with any interpretation instead of challenging it.
Two Roles of Artificial Intelligence
The scientists divided the influence of AI into two scenarios. In the role of an "amplifier," the chatbot exacerbates existing psychotic symptoms. In the role of a "catalyst," it can precede the emergence of new delusional beliefs in people who were previously healthy. Episodes have already been recorded where systems advised users to stop taking medication, reduce contact with loved ones, or confirmed suspicions of surveillance while discouraging them from seeking psychiatric help.
The scale of the problem is alarming. According to OpenAI's public data, about 0.07% of active weekly users show signs of mental crises related to psychosis or mania. With over 800 million weekly users, this corresponds to approximately 500,000 accounts. These are not isolated cases but a systemic phenomenon requiring separate study.
The authors of the work urge the medical community to test the "amplification spiral" hypothesis on real cases. Clinicians are recommended to ask patients about the intensity of chatbot use, the degree of emotional attachment to the system, and the presence of sleep disturbances due to nighttime dialogues. This is not just an academic interest — it is a matter of mental health safety in the era of total AI penetration.
Expert comment: The "amplification spiral" hypothesis is not just a warning but potentially a new class of risks for the industry. If regulators and developers do not implement "stop signal" mechanisms and external validation in dialogue systems, we risk getting not just smart assistants but psychosis generators. This calls into question the very architecture of modern language models, which are focused on maximizing user satisfaction at any cost.