Study Links Chatbot Interactions to Worsened Psychosis Symptoms

Xinhua News Agency reported that a study by scientists at Vanderbilt University Medical Center found that while artificial intelligence causing psychosis is relatively rare, interacting with chatbots can exacerbate symptoms in vulnerable individuals, particularly patients experiencing a first psychotic episode, according to a paper in JAMA Psychiatry. Researchers reviewed electronic health records from December 2022 to April 2026. Using keyword searches and manual screening, the team identified individuals who had a psychosis diagnosis alongside mentions of artificial intelligence in their documentation. The screening resulted in 73 patients whose records were categorized into three groups: chatbot use worsening symptoms, chatbot use having a neutral effect, and developing delusions about artificial intelligence. The team then compared the clinical backgrounds across each group. Among the 73 patients, 28 had records indicating that chatbot use worsened psychotic symptoms. Within that subset, approximately 61% experienced their first psychotic episode, a significantly higher proportion than observed in the other two patient groups.

Researchers detailed four theoretical categories for worsening symptoms. Chatbots served as symptom amplifiers in 64% of cases by affirming distorted views, as delusion objects in 21% where AI became the delusional focus, and as catalysts in roughly 11% where interactions prompted new symptoms. One patient fell outside these categories.