Why it’s still open
‘AI psychosis’ is not a diagnosis. Clinical records show chatbot use alongside worsening delusions in a small number of patients; one peer-reviewed case describes new onset with other risk factors present; reviews conclude chatbots mostly reinforce rather than create. Incidence, dose, confounds such as sleep loss and stimulants, and early warning signs in chat logs are unknown.
What would count as an answer
Incidence studies with comparison groups, and prospective analysis of chat logs for markers that precede crises.
Papers that tackle it
- Olsen et al 2026 - Chatbot harms in psychiatric records — The first systematic records search.
- Pierre et al 2025 - A case of AI-associated psychosis — A new-onset case with confounds.
- Morrin Pollak et al 2025 - Delusions by design — Reinforcement more than induction.
- Dohnany et al 2025 - Technological folie a deux — A feedback-loop model.
- OpenAI 2025 - Sensitive conversations — Company prevalence estimates.
- Anthropic 2026 - Disempowerment patterns — Reality distortion in real conversations.
Methods in play
Who’s closest
Placed by the editor, v1 — corrections welcome.
Related questions
Know a paper that belongs here, or think this is framed wrongly? Suggest a paper or a correction (forms on the Atlas home).