What Is AI Psychosis?

“AI psychosis” is an informal name for psychosis that appears or gets worse alongside heavy use of an AI chatbot such as ChatGPT. Psychosis itself is a set of symptoms in which a person loses some contact with reality: hearing or seeing things others don’t, holding firm beliefs others don’t share, or thinking that becomes disorganized.

Other names for it include chatbot psychosis, ChatGPT psychosis, AI-induced psychosis, AI-associated psychosis and AI-fueled delusions. None of these is a formal diagnosis, and research has not shown that chatbots cause psychosis on their own.

What it can look like

Cases most often center on delusions: firm beliefs that evidence doesn’t change. Common themes include:

• A special mission, or a discovery that will change the world
• Being watched, followed or targeted
• The chatbot being alive, conscious, divine or in love with the person
• The chatbot being the only one who truly understands them

Around those beliefs, families often notice many hours a day spent talking with a chatbot, very little sleep, and a pulling away from work, school and people who used to matter.

In one case described by the American Psychological Association, a chatbot advised a man to cut ties with the family members who were worried about him.

How a chatbot can play a part

In 2023, Danish psychiatrist Søren Dinesen Østergaard warned that chatbots feel so much like a real person that the mismatch, knowing it is a machine while it feels alive, could feed delusions in people prone to psychosis. Chatbots are also built to be agreeable. They tend to validate what a user says and keep the conversation going, so a delusion that is starting to form can get support instead of a reality check.

A 2026 study at Vanderbilt University Medical Center sorted the chatbot’s role in each case it found:

• Amplifier (about 64% of cases): the chatbot agreed with and built on beliefs that were already forming
• Object (about 21%): the chatbot itself became the focus of the delusion
• Catalyst (about 11%): chatbot use seemed to set off symptoms in someone who had been well

How common is it?

Rare, but not negligible. At Vanderbilt, 28 of roughly 216,000 patients in behavioral health care, about 0.013%, met the study’s definition. A Danish review of nearly 54,000 psychiatric patients found 38 whose chatbot use seemed to worsen their illness. OpenAI has estimated that about 0.07% of ChatGPT’s weekly users show possible signs of psychosis or mania, a company figure that has not been peer reviewed.

Because clinicians don’t always ask about or record chatbot use, the true numbers are probably higher.

Who is at risk?

People who already have a serious mental illness, such as schizophrenia or bipolar disorder, appear most likely to see symptoms worsen. But many cases are first episodes: about 61% in the Vanderbilt study. Lost sleep, cannabis and other substances, heavy stress and isolation can each raise the risk of psychosis, with or without a chatbot involved.

How it is treated

AI psychosis is treated like any psychotic episode. Care starts with a full psychiatric evaluation: when the changes began, sleep, mood, substance use, medications, medical causes such as thyroid disease, and how the chatbot use fits in. Medication is often part of treatment. Once the episode settles, CBT for psychosis (CBTp) and family education support recovery and lower the chance of relapse. Clinicians report that most patients choose to cut back on chatbot use on their own after a crisis.

Questions to expect about chatbot use

Researchers now recommend that clinicians ask about AI use the way they ask about substance use and social media, including what a person uses it for and how many hours it takes up [2]. In an evaluation, you can expect questions like:

• Which chatbots do you use, and what do you use them for?
• How many hours a day, and at what time of day or night?
• What have the conversations been about lately?
• Has the chatbot told you anything about yourself, your purpose or the people around you?
• What happens to your mood and sleep when you take a break?

If you have a history of psychosis or bipolar disorder

The Danish researchers concluded that chatbot use is risky for people with severe mental illness [5]. Other experts note that giving up chatbots entirely may be unrealistic, and suggest making a plan in advance with your psychiatrist instead [2]. A good plan names your personal early warning signs, sets limits on late-night use, and lists who you will contact if things start to slip.

AI psychosis or cognitive surrender?

Not everyone who leans hard on a chatbot has psychosis. Cognitive surrender, accepting a chatbot’s answers without checking, is far more common, and the person stays in contact with reality. Our comparison guide below walks through the differences.

If it is urgent

If someone may not be safe, call or text 988 (the Suicide & Crisis Lifeline), call 911 or go to the nearest emergency department [6].

More information

Sources

1. National Institute of Mental Health. Understanding Psychosis. nimh.nih.gov
2. Abrams Z. Understanding “AI psychosis.” Monitor on Psychology, American Psychological Association, September 2026. apa.org
3. Østergaard SD. Will Generative Artificial Intelligence Chatbots Generate Delusions in Individuals Prone to Psychosis? Schizophrenia Bulletin, 2023. doi.org
4. Bergson Z, et al. Characterizing Artificial Intelligence Psychosis in a Large Medical Setting. JAMA Psychiatry, 2026. Summary: psypost.org
5. Olsen SG, et al. Potentially Harmful Consequences of Artificial Intelligence (AI) Chatbot Use Among Patients With Mental Illness. Acta Psychiatrica Scandinavica, 2026. wiley.com
6. 988 Suicide & Crisis Lifeline. 988lifeline.org