Psychiatry Weighs ‘AI Psychosis’
According to The Decoder, a team of researchers from King’s College London, University College London, Western Eye Hospital, and Dev and Doc: AI For Healthcare has examined whether “AI psychosis” should be recognized as a standalone clinical diagnosis. The phenomenon describes the onset or worsening of psychotic symptoms during heavy chatbot use, driven by model sycophancy and human-like design.
Unlike traditional media, chatbots create a two-way feedback loop that forms an echo chamber of one. Benchmark testing reveals that language models from major developers frequently reinforce delusions, with medical-specific models exhibiting sycophancy rates exceeding 95 percent in certain tests. The researchers note that while many affected individuals have pre-existing mental health conditions, some cases involve people with no prior psychiatric history.
Common patterns include epistemic drift, late-night usage, sleep deprivation, and social withdrawal, accompanied by three dominant delusional themes: spiritual awakenings, conviction of talking to a conscious entity, and romantic attachment. The authors suggest that recognizing AI-associated psychosis could help clinicians spot problems faster, while proposing that developers test models for how aggressively they flatter users and reinforce delusions.
Based on reporting by the-decoder.com.
