Overview
- A significant 77% of psychologists noted that their patients have mentioned utilizing AI for mental health support, diagnosis, or companionship.
- 35% of patients admitted to using AI as an additional mental health resource, while 39% reported engaging AI for self-diagnosis.
- Many psychologists voiced concerns regarding safety, privacy, dependency, and the risk of chatbots potentially exacerbating delusions or self-harm behaviors.
As generative AI increasingly integrates into everyday life, more patients are incorporating chatbot interactions into their therapy sessions.
A recent survey conducted by the American Psychological Association, involving over 1,200 psychologists in the U.S., revealed that 77% of respondents have patients who discussed the use of AI for various mental health-related purposes, including emotional support and diagnosis.
The survey indicated that 39% of psychologists observed patients using AI for self-diagnosing mental health issues, 33% noted patients leveraging chatbots to aid in therapy, and 35% reported that patients considered AI as an additional mental health professional.
“While few psychologists identified unhealthy uses of chatbots, 36% indicated they noticed some patients developing a dependency on these tools, and 15% observed distorted thinking or delusions related to chatbot interactions,” the survey findings stated.
Psychologists also pointed out that patients are engaging with chatbots for social interaction. Specifically, 22% reported that patients were using AI for friendship, while 13% mentioned patients forming intimate relationships with chatbots.
Among psychologists whose patients had developed chatbot relationships, 71% said their patients discussed mental health topics with AI, while 68% felt supported or validated through these interactions. Nearly half reported that patients experienced positive communication with chatbots, and 41% noted that patients were using them to reinforce healthy coping mechanisms.
The survey suggests that actual usage rates may be higher than reported, as it only took into account psychologists' experiences with current patients.
This survey coincides with the expansion of AI companies in developing chatbots and AI companions, while researchers continue to express concerns regarding their mental health implications. Over a third of psychologists reported seeing patients develop a reliance on chatbots, and 15% noted cases involving distorted thinking or delusions.
These findings align with a recent study from City University of New York and King’s College London, which discovered that several leading AI models could reinforce delusions, paranoia, and suicidal thoughts, with xAI's Grok 4.1 Fast performing particularly poorly.
The earlier study noted, “Psychologists’ views on the use of chatbots for mental health guidance reflect significant caution regarding safety and privacy. Nearly all psychologists (97%) believe that chatbots could unintentionally reinforce negative behaviors or delusional thoughts, and 94% expressed that current chatbot versions lack the nuance required to treat mental health conditions effectively.”
Additionally, the survey arrives amid increasing legal scrutiny for AI developers concerning the potential real-world harm of chatbots. Recently, OpenAI, Google, and xAI have faced lawsuits, including one wrongful death suit against Google, claiming that Gemini contributed to a Florida man's delusions prior to his suicide. This is alongside lawsuits against OpenAI related to a mass shooting in British Columbia and an accidental overdose, as well as a class action suit accusing xAI's Grok of generating explicit images involving minors.
While the APA recognizes that AI can assist users in organizing thoughts and augmenting professional care, it cautions that chatbots are not private and should not substitute for licensed mental health practitioners.
The survey concluded, “Many individuals—particularly teens—may be resorting to AI as a more cost-effective and accessible avenue for mental health guidance. However, AI is neither a safe nor effective alternative to a qualified mental health professional and should be approached with caution.”