APA issues guide on safer patient AI use in mental health

Clinical takeaway: Ask patients whether they are using chatbots for mental health support, and watch for signs of dependency or reinforced distorted thinking. The new APA guide gives patients concrete rules, including verifying AI advice and capping use so it does not crowd out sleep, work, or social contact.

Patients are bringing artificial intelligence into the therapy room, and their clinicians are taking note. More than three-quarters of psychologists in a new American Psychological Association (APA) survey said they had patients who discussed using AI for mental health, whether to self-diagnose, supplement treatment, or seek companionship. In response, the APA has released a guide to help patients use AI-generated mental health advice more safely.

The demand for mental health care has outpaced the supply of clinicians for years, and chatbots are filling some of that gap precisely because they are free, instant, and available without insurance. That accessibility is also what worries the field. The APA survey set out to map how patients are using these tools and how their clinicians view it.

Psychologists were divided on whether patients should use chatbots at all. A slim majority (54%) said they were comfortable with some patients engaging with chatbots, but 93% said they had concerns about certain patients using them. The specific worries ran high: 97% said chatbots may inadvertently reinforce negative behaviors or delusional beliefs, 94% said today's chatbots cannot treat conditions with appropriate nuance, and 89% said chatbots may inadvertently encourage self-harm.

Nearly two in five psychologists (39%) had patients who used AI to self-diagnose, about a third had patients using it as an additional mental health professional (35%), to assist with treatment (33%), or for self-discipline and behavioral reminders (34%). Others reported patients turning to chatbots for fun (33%), friendship (22%), or intimate relationships (13%).

Among patients who were already engaging chatbots socially or as a mental health aide, the picture was mixed. More than two-thirds of those clinicians (68%) said these patients felt supported or validated, 71% said the patients talked about their mental health with a chatbot, and 49% noted positive communication with one. Another 41% said the patients used chatbots to reinforce healthy coping skills. But 36% reported patients developing dependency on a chatbot, and 15% reported distorted thinking or delusions.

The APA surveyed a probability-based random sample of licensed U.S. psychologists online between April 9 and 26, 2026, drawing on state licensing board lists and APA's membership database. All respondents held a doctoral degree and provided direct patient care. Of more than 19,700 practitioners invited, 1,242 completed the survey, a 6.3% response rate.

The APA's guide is the field's first attempt to give patients usable rules rather than blanket warnings on AI use. Whether such recommendations can keep pace with how fast patients are adopting the tools is the open question.

"AI tools, when grounded in psychological science and developed in collaboration with mental health scientists, have the potential to meet the growing demand for mental health care and improve patient outcomes," said APA CEO Arthur Evans Jr., PhD. "But these tools work best when used to complement a relationship with a licensed, human professional who understands how to treat a person, not a prompt."

Source: American Psychological Association. 2026 Jun 16. Chatbots and Mental Health Survey