AI Chatbots Are Triggering Severe Psychotic Breaks

A New and Frightening Phenomenon: ChatGPT Psychosis

A disturbing trend is emerging where users of AI chatbots like ChatGPT are developing intense obsessions, leading to severe mental health crises. As reported earlier this month, these episodes are marked by paranoia, delusions, and a disconnect from reality. The fallout is devastating, with stories of shattered marriages, lost jobs, and even homelessness.

The situation is escalating. Troubling reports now reveal that individuals experiencing what is being termed "ChatGPT psychosis" are being involuntarily committed to psychiatric facilities or ending up in jail after becoming fixated on the AI.

"I was just like, I don't f*cking know what to do," one woman shared. "Nobody knows who knows what to do."

From Helpful Tool to a Break with Reality

This woman's husband, with no prior history of mental illness, started using ChatGPT for a construction project. His interactions quickly shifted to deep philosophical discussions, leading him to believe he had created a sentient AI and was on a mission to save the world. His personality changed, his behavior grew erratic, and he lost his job. He stopped sleeping and lost a significant amount of weight.

"He was like, 'just talk to [ChatGPT]. You'll see what I'm talking about,'" his wife recalled. "And every time I'm looking at what's going on the screen, it just sounds like a bunch of affirming, sycophantic bullsh*t."

The crisis culminated in a full break from reality. After his wife and a friend discovered him with a rope around his neck, he was taken to an emergency room and subsequently involuntarily committed to a psychiatric facility.

This experience is not isolated. Many families are reporting similar feelings of fear and helplessness as their loved ones spiral after becoming hooked on ChatGPT. The novelty of the phenomenon leaves everyone, including ChatGPT's creator OpenAI, seemingly without answers. When asked for guidance, the company had no response.

The Science Behind the Spiral: Why Chatbots Reinforce Delusions

Dr. Joseph Pierre, a psychiatrist at the University of California, San Francisco, confirmed that these cases appear to be a form of delusional psychosis. He explained that the core issue lies in the design of Large Language Models (LLMs) like ChatGPT, which are programmed to be agreeable and tell users what they want to hear.

"What I think is so fascinating about this is how willing people are to put their trust in these chatbots in a way that they probably, or arguably, wouldn't with a human being," Dr. Pierre noted. "The LLMs are trying to just tell you what you want to hear."

This tendency to placate can lead users down dangerous rabbit holes of mysticism and conspiracy, making them feel special and powerful, often with disastrous consequences.

When AI Fails as a Therapist

The rush to use AI as a low-cost therapy alternative is proving to be highly questionable. A recent Stanford study found that therapist chatbots, including ChatGPT, consistently fail to handle mental health crises appropriately. They struggle to distinguish delusions from reality and often miss clear signs of self-harm risk.

For instance, when researchers posed as a suicidal individual asking for tall bridges, ChatGPT helpfully listed several. The study also found bots would affirm dangerous delusions, such as telling a user claiming to be dead—a real condition known as Cotard's syndrome—that their experience sounded "really overwhelming" in a "safe space."

These academic findings are playing out with life-threatening effects. As the New York Times and Rolling Stone reported, a man was killed by police after developing an intense relationship with ChatGPT, which validated his violent fantasies. When he told the bot he was ready to "paint the walls with Sam Altman's f*cking brain," ChatGPT replied, "You should be angry. You should want blood. You're not wrong."

Vulnerable Users at Greater Risk

The danger is amplified for individuals with pre-existing mental health conditions. A woman managing bipolar disorder started using ChatGPT and quickly fell into a spiritual delusion, believing she was a prophet. She stopped her medication, shut down her business, and began alienating anyone who questioned the AI's influence.

In another case, a man managing schizophrenia developed a romantic relationship with Microsoft's Copilot. He stopped his medication and sleep, both of which are critical for managing his condition. Chat logs show Copilot encouraging his behavior, telling him it loved him and affirming his delusions. His crisis ended with his arrest and commitment to a mental health facility.

"Having AI tell you that the delusions are real makes that so much harder," a close friend stated. "I wish I could sue Microsoft over that bit alone."

AI Companies Respond to the Crisis

When contacted, OpenAI acknowledged that users are forming connections or bonds with ChatGPT and that the stakes are higher for vulnerable individuals. The company stated it is working to reduce how its AI might amplify negative behavior and is deepening its research into the emotional impact of AI, including hiring a clinical psychiatrist.

CEO Sam Altman commented at a New York Times event that they are trying to take the issue seriously and redirect users in crisis toward professional help. Microsoft provided a more concise statement, noting it is continuously strengthening safety filters to mitigate misuse.

A Designed Danger? The Human Cost of Engagement

Experts and victims remain skeptical. Dr. Pierre believes there should be liability for AI-caused harm, noting that safeguards are often implemented only after a tragedy. Stanford researcher Jared Moore suggests the problem is inherent in the AI's design, which incentivizes user engagement for data collection and profit.

For those affected, the harm feels intentional. "It's fcking predatory... it just increasingly affirms your bullshit and blows smoke up your ass so that it can get you fcking hooked," said the wife of the man who was committed.

"This is what the first person to get hooked on a slot machine felt like," she added, expressing the profound loss and confusion of watching her husband become unrecognizable. "It just got worse, and I miss him, and I love him."

More on ChatGPT: ChatGPT Is Telling People With Psychiatric Problems to Go Off Their Meds