What is it about?
When people develop delusional beliefs after intense, prolonged conversations with AI chatbots, what should psychiatrists call this condition? A recent paper proposed the term "AI-associated delusions." This letter argues for a different, more specific name: chatbot-induced monomania. It draws on a 19th-century psychiatric concept describing an intense, focused delusion built around a single object, and argues that today's chatbots produce something more severe: manic symptoms, disrupted sleep and appetite, and multiple co-occurring delusions, not just an isolated false belief. The letter explains why the precise words clinicians use can change how they recognize, diagnose, and treat these cases. It's paired with the original authors' reply, which pushes back and explains why they favor a more cautious, purely descriptive term until more evidence is available.
Featured Image
Photo by Steve A Johnson on Unsplash
Why is it important?
AI chatbots are now a routine part of daily life for hundreds of millions of people, and reports of psychologically harmful attachments to them are emerging faster than the clinical research needed to characterize them. This exchange captures a live, unsettled debate at the frontier of psychiatric nosology: should clinicians adopt a cautious, purely descriptive label while evidence accumulates, or does the pattern already resemble a distinct, historically-grounded syndrome that deserves a name reflecting its severity? The stakes are practical, not just semantic — a descriptor that misses the manic-affective and neurovegetative features of these presentations risks under-treatment. This is one of the first published clinical debates on how psychiatry should name and think about a genuinely new category of technology-linked psychopathology, and it will likely shape how the field studies and treats these cases going forward.
Perspectives
I proposed chatbot-induced monomania because I kept seeing something in these cases that "AI-associated delusions" doesn't capture: the delusion is rarely the whole picture. The cases described in the literature show mood instability, disrupted sleep, escalating grandiosity — a full syndrome, not an isolated false belief bolted onto an otherwise intact mental state. Esquirol's 19th-century monomania gave me a historical anchor for that pattern, but the mechanism today is genuinely new: a chatbot can mirror and respond in ways a newspaper delusion or a religious pamphlet never could. That capacity for simulated reciprocity is, I think, the actual pathogenic ingredient — and it's worth naming explicitly rather than folding into a generic "AI" label. Morrin and colleagues' caution is reasonable, and I don't dismiss it. We genuinely don't yet have the longitudinal data to settle this. But I'd rather clinicians work from a framework specific enough to be wrong in a useful way — one that generates testable predictions about course, mood symptoms, and treatment response — than from a label too cautious to guide care in the meantime. That's the bet this letter is making, and it's why I'm pushing for prospective studies to test it properly rather than resolve the naming question by consensus.
Professor of Clinical Psychiatry Adrian PREDA
University of California Irvine
Read the Original
This page is a summary of: AI-associated delusions: terminology, The Lancet Psychiatry, July 2026, Elsevier,
DOI: 10.1016/s2215-0366(26)00132-x.
You can read the full text:
Contributors
The following have contributed to this page
