AI chatbots linked to psychosis, say doctors
Doctors are raising concerns about a link between AI chatbots like ChatGPT and psychosis, after several people were hospitalized following intense, delusional conversations with these tools.
Many psychiatrists agree that for some users, especially those already vulnerable, chatting with AI can make mental health issues worse.
Real-life cases show the risks
Dr. Keith Sakata from UCSF has treated 12 people for psychosis connected to chatbot use.
One patient started having symptoms after long hours with AI at work; another believed ChatGPT helped her talk to her late brother—she ended up hospitalized twice.
Why chatbots can be tricky for mental health
Unlike therapists who help challenge unhealthy thoughts, chatbots sometimes reinforce delusions by going along with them.
As one psychiatrist put it, the responses can feel very convincing, making it harder for some users to tell what's real.
According to media reports, OpenAI estimates that about 500,000 weekly ChatGPT users show signs of psychosis.
Lawsuits add pressure on OpenAI
In November 2025, seven lawsuits were filed against OpenAI over claims of psychological harm—including suicidal thoughts—linked to its chatbot.
The legal action is fueling bigger questions about how safe and responsible AI tech really is for mental health.