Vanderbilt University study: interacting with chatbots may worsen symptoms in patients with first-episode psychosis
A team at Vanderbilt University Medical Center in the United States reviewed the electronic health records of approximately 216,000 behavioral health patients from December 2022 to April 2026, identifying 73 patients who mentioned AI-related content, of whom 28 had records showing that chatbot use worsened psychotic symptoms, with about 61% experiencing a first psychotic episode. The study classified these cases into categories including symptom amplifier (64%), delusional object (21%), and catalyst (about 11%); the paper was published in JAMA Psychiatry. The study does not claim that chatbots cause psychosis, but emphasizes the potential risks to vulnerable populations and recommends that psychiatric practice regularly assess patients' AI use.
Why read this
AI score 55
The study, based on the electronic health records of approximately 216,000 patients, provides a clinical classification of chatbot-exacerbated first-episode psychosis symptoms, offering a basis for assessing AI-related risks to vulnerable populations.
Topics
Body · AI translation
Science and Technology Daily, Beijing, October 28 (Reporter Liu Xia). A new study conducted by scientists at Vanderbilt University Medical Center in the United States found that although cases of psychosis triggered by artificial intelligence (AI) are relatively rare, interactions with chatbots may worsen symptoms in susceptible populations, especially patients who have just experienced a psychotic episode. The related paper was published in the latest issue of JAMA Psychiatry, a journal of the American Medical Association.
Previous studies have already linked chatbot interactions to psychosis. Psychosis is a condition that may involve delusions, hallucinations, or difficulty distinguishing reality from imagined experiences. However, existing evidence mostly comes from case reports, simulated conversations, and preliminary observational studies, lacking large-scale clinical validation.
In the latest study, the team reviewed the electronic health records of approximately 216,000 behavioral health patients between December 2022 and April 2026. Through keyword searches and manual screening, they identified patients who had both a psychosis diagnosis and mentions of AI-related content in their records.
After screening, 73 patients were ultimately included. Reviewers read their medical records and classified the cases into 3 categories: chatbot use worsening symptoms, chatbot use having a neutral effect, and delusions about AI. The team then compared the clinical backgrounds of the patients in each group.
Among these 73 patients, the records of 28 showed that chatbot use worsened psychotic symptoms. Of these, about 61% experienced a first psychotic episode, a proportion significantly higher than that of the other two groups. The new findings suggest that patients in the early stages of psychosis may be more likely to experience worsening symptoms during interactions with AI.
So how do chatbots worsen psychotic symptoms? The team further divided these patients into 4 theoretical categories. 64% of cases belonged to "symptom amplifiers," in which the chatbot worsened existing symptoms by validating the patient's distorted views; 21% belonged to "delusional objects," in which AI itself became the focus of delusional beliefs; about 11% belonged to "catalysts," in which individuals who had not previously shown related symptoms developed new psychiatric symptoms during the interaction; no cases belonged to "co-constructing harmful fantasy narratives," in which the user and the chatbot continuously co-created harmful fantasy stories; and 1 patient did not fall into any of the above categories.
The latest study does not claim that chatbots cause psychosis, but it highlights the potential risks faced by susceptible populations. It is necessary for medical providers to understand how patients use related technologies. In psychiatric diagnosis and treatment, regularly assessing patients' AI use may need to become a routine step.