If chatbots seem to trigger someone's psychosis, do doctors need a brand-new diagnosis, or does the existing rulebook already cover it?

This explores whether AI-related mental health problems need new diagnostic labels, or whether they already fit inside established frameworks like ICD-11. The corpus answers this directly in one note and only sideways in a few others, and it never walks through specific ICD-11 codes.

The direct answer is a perspective article arguing against recognizing "AI-associated psychosis" as a new disorder. It proposes treating AI use as an environmental stressor inside existing psychosis formulations. The diagnosis stays the same, and AI becomes part of the story of why the episode happened when it did and what it was about Should we recognize AI-associated psychosis as a new disorder?. The argument is partly about evidence: the data is thin and causation can't yet be attributed to AI. On that view, a new category would lock in a claim nobody can support yet.

The harder test is that AI-related harm isn't one thing. One note traces a single move, treating a system as a mind, into several separate risks: emotional dependence, erosion of autonomy, erosion of status, and political conflict Does perceiving AI as conscious create multiple distinct risks?. It doesn't map these to diagnoses. Some, like dependence, sound clinical. Others, like status erosion, sound social or economic. So the stressor framing may cover the psychosis case well while leaving open which of the other risks belong to clinical categories at all. The same note argues that interaction-design fixes aimed at that perceptual move work more directly than system-level alignment. That suggests the lever may sit upstream of any diagnostic label.

Another note offers a possible reason the stressor framing fits. It argues that AI output is "event-residue" rather than real utterances, and that the meaning of the exchange is supplied by the human, who animates the text into something that feels like a conversation Does AI generate genuine utterances or just text patterns?. This is my inference, not the note's claim. If the exchange's structure lives on the human side, then existing frameworks, which already describe how people build meaning and beliefs, may cover the mechanism. AI would supply the material rather than a new mechanism.

The collection is thin here. It has one note that takes a position on diagnostic categories, and nothing on how ICD-11's specific chapters, such as psychotic, mood, or dependence-related disorders, would apply to AI presentations. If you want to pursue this, the psychosis note is the doorway, and the consciousness-attribution note shows why one label probably won't cover everything.