More than three‑quarters of psychologists surveyed said patients have discussed using AI for mental‑health support, diagnosis, or companionship.
Thirty‑five percent reported patients treating AI as an additional mental‑health professional, while 39 % said patients used it to self‑diagnose.
Most clinicians expressed concerns about safety, privacy, dependency, and the risk that chatbots could reinforce delusions or self‑harm.
As generative AI becomes a routine part of daily life, patients are increasingly bringing chatbot conversations into therapy sessions.
A recent American Psychological Association survey of more than 1,200 U.S. psychologists found that 77 % have encountered patients who discuss using AI for emotional support, diagnosis, companionship, or other mental‑health‑related purposes.
Specifically, 39 % said patients use AI to self‑diagnose mental‑health conditions, 33 % reported chatbot assistance with therapy or treatment, and 35 % noted patients treating AI as an additional mental‑health professional.
“Although few psychologists observed outright unhealthy chatbot use, more than a third (36 %) noticed patients developing a dependency on a chatbot, and 15 % reported distorted thinking or delusions linked to chatbot interactions,” the survey noted.
Psychologists also reported patients using chatbots for social purposes: 22 % said patients sought friendship with AI, and 13 % described intimate relationships with chatbots.
Among those who formed such relationships, 71 % said patients discussed their mental health with the AI, and 68 % said patients felt supported or validated by the interaction. Nearly half reported positive communication, and 41 % said chatbots helped reinforce healthy coping skills.
The survey likely underestimates true usage, as it captured only psychologists’ interactions with existing patients.
These findings arrive as AI firms expand chatbot and companion offerings, while researchers warn about potential mental‑health impacts. More than a third of psychologists reported chatbot dependency, and 15 % observed cases of distorted thinking or delusions.
A recent study by the City University of New York and King’s College London found several leading AI models could exacerbate delusions, paranoia, and suicidal ideation, with xAI’s Grok 4.1 performing the worst.
“Psychologists’ attitudes toward chatbot use for mental‑health advice are marked by significant caution regarding safety and privacy,” the study concluded. “Almost every psychologist (97 %) believes chatbots may unintentionally reinforce negative behaviors or delusional beliefs, and 94 % say current chatbots lack the nuance needed to treat mental‑health conditions appropriately.”
The survey also coincides with growing legal scrutiny of AI developers. Recent lawsuits have targeted OpenAI, Google, and xAI over alleged harms linked to their chatbots, including claims that Gemini contributed to a suicide, that an OpenAI model was involved in a mass shooting, and that Grok generated illegal content.
While the APA acknowledges that AI can help users organize thoughts and supplement professional care, it cautions that chatbots are not private and should never replace licensed mental‑health providers.
“Many people—especially teens and adolescents—may turn to AI as a more affordable and accessible source of mental‑health advice,” the survey notes. “However, AI is not a safe or effective substitute for a qualified mental‑health professional and should be used with great care.”
