AI psychosis in the health record: rare but real, first-episode patients the most vulnerable
Where it stands: A study based on a review of electronic health records at a large medical center suggests psychosis linked to conversational AI chatbots is rare but real, and that first-episode patients may be the most vulnerable group. The available account does not include the originating outlet, journal, case counts, or whether a causal role was established. It remains to be seen what the primary publication and dataset show, whether causation is supported, and whether clinicians or regulators issue guidance for first-episode patients.
2026-10-23 - 2026-10-23 (~8:00 AM ET, MSN syndication — msn.com/en-us): a study built on a review of electronic health records at a large medical center suggests conversational-AI-chatbot-linked psychosis appears rare but real, with first-episode patients possibly the most vulnerable group. Syndication-grade capture: origin outlet, journal, case counts, and causal framing not in it. Watch: the primary publication and dataset detail, whether any causal role is established, clinician or regulator guidance for first-episode patients.