Conversational artificial intelligence (AI) is being used for emotional support by adults and adolescents alike. These systems (for example, ‘chatbots’) prioritize continued user engagement and therefore are more prone to validate users than to challenge them. This built-in characteristic might present a problem for people with psychotic vulnerability who look for emotional support in conversational AI when coping with delusional beliefs.
In their 2025 paper, Hudon and Stip examined how the dynamics of conversational AI might trigger, amplify or reshape psychotic experiences in vulnerable individuals. The paper’s background theory is the stress-vulnerability framework, according to which vulnerabilities (such as genetic predisposition, childhood trauma, sleep disruption, social isolation and problematic cognitive styles) interact with environmental stressors and allostatic load (the wear and tear on the body after repetitive adaptive responses to stressors) to lower the threshold for psychosis onset. This is a preview of subscription content, access via your institution Receive 12 digital issues and online access to articles Prices may be subject to local taxes which are calculated during checkout Hudon, A. & Stip, E.
Delusional experiences emerging from AI chatbot interactions or “AI psychosis”. Health 12, e85799 https://doi.org/10.2196/85799 (2025) Article PubMed PubMed Central Google Scholar Myin-Germeys, I. & van Os, J. Stress-reactivity in psychosis: evidence for an affective pathway to psychosis.
Rev. 27, 409–424 (2007) Department of Psychology, Lund University, Lund, Sweden The author declares no competing interests. When conversational AI validates delusional beliefs. Nat Rev Psychol (2026). https://doi.org/10.1038/s44159-026-00620-8 DOI: https://doi.org/10.1038/s44159-026-00620-8
Extract — continue reading at the source.
