There is no clinical or research evidence to support its existence.

AI psychosis is also not a recognized clinical diagnosis. But there is a small and growing body of literature showing that people with no history of psychosis can develop delusional thinking during immersive, extended chatbot use.

The British Journal of Psychiatry has published a framework on harm reduction. Psychiatric Times has urged clinicians to ask about AI use at intake and document it like substance use.

Anti-AI psychosis has none of that underneath it. No case reports, no clinicians describing a pattern in patients. What it has is the word “psychosis,” borrowed from a term that earned it, so the two sound like a matched pair. That borrowed symmetry is a trick.

I know some of this because my wife is a clinical psychiatrist in a tech-saturated regions of the country. I know the rest because I work at Microsoft AI, designing ads experiences for AI search products. A lot of my job is arguing for a version where the ad doesn’t hijack the conversation, but that I have to make that argument is the point. When someone says AI is going to be monetized irresponsibly, I don’t hear psychosis. I hear the fight I’m already in day-to-day.

Don’t let Designer/Builder AI boosters weaponize the term. Whatever the intent, the effect is twofold: it trivializes people living with actual symptoms of psychosis, and it lets boosters skip the argument and diagnose the person instead.

Job displacement, anger at how the tools are being built, predatory monetization: these are valid claims, and they deserve answers. “You’re psychotic” is not one.