Therapists’ Concerns About AI: What the 2026 APA Survey Shows
Most therapists are not reflexively anti-technology, and the survey data says so. In the American Psychological Association’s 2026 Chatbots and Mental Health Survey of 1,242 licensed psychologists, 93% reported concerns about certain patients using AI chatbots, while 54% said they were comfortable with some patients using them. That split is the honest starting point for this conversation, and it is where we want to begin.
What do therapists actually say about AI in therapy?
APA surveyed 1,242 licensed psychologists in April 2026. More than three in four said their patients are discussing AI in therapy. Among the concerns reported: 97% said chatbots may inadvertently reinforce negative behaviors or delusional beliefs, 94% said today’s chatbots cannot treat conditions with an appropriate amount of nuance, and 89% said chatbots may inadvertently encourage self-harm.
Clinicians also reported what they were seeing in the room: 36% said patients were developing a level of dependency on a chatbot, and 15% said patients were developing distorted thinking or delusions. Thirty-nine percent had patients who used AI to self-diagnose, and 35% had patients using it as an additional mental health professional.
The picture is not uniformly negative, and we think it matters that we say so. Two in five psychologists (40%) were optimistic that chatbots could help patients when a mental health professional is not available, and only 24% believed patients will one day prefer therapy chatbots to human clinicians.
Is the worry about AI helping, or AI replacing?
Replacing. APA’s November 2025 health advisory is specific on this point: these tools may be appropriate as a supportive adjunct, not a substitute, to an ongoing therapeutic relationship. Nobody is upset about scheduling software. What concerns us is a chatbot positioned as the first line of treatment, with a licensed human as an escalation tier you reach only after deteriorating enough to qualify for one.
APA chief executive Arthur C. Evans Jr. described the appeal plainly in June 2026: accessible chatbots offer the path of least resistance because they are supportive to a fault, readily available, and easy to access without insurance. That final phrase is the economic heart of the matter.
Why does insurance make this worse?
Because the access problem in mental healthcare is already an economic one. A 2024 study by health economists at RTI International, using commercial insurance claims data, found that patients were 10.6 times more likely to go out of network to find psychological care than to find specialty medical care, and identified low reimbursement for mental health providers as a major cause of inadequate networks. APA summarized the conclusion bluntly: the access problem is not due to a shortage of providers, but a shortage of in-network providers.
Availability is strained too. As of December 31, 2025, federally designated mental health shortage areas covered roughly 137 million people in the United States and met about 27% of estimated need. In New York State that figure was closer to 15%, according to KFF State Health Facts.
Put those two facts together and the risk becomes clear. If a payer can point to a free digital tool as adequate care, it becomes easier to deny sessions, cap the length of treatment, and refuse rate increases inside a system that already reimburses behavioral health less favorably than medical and surgical care.
What happens to clinical training and the next generation of therapists?
This part is our concern rather than an established finding, and we want to label it that way. We are not aware of data measuring it yet. Intake, screening, documentation, and lower-acuity support are the roles where new clinicians accumulate supervised hours toward licensure. If those rungs are automated, we expect the ladder into the profession to narrow, and fewer licensed clinicians five years from now means longer waits, not shorter ones.
The field appears to see the training question as live. APA’s advisory calls on graduate programs and clinical training directors to make sure trainees learn about these technologies and how to evaluate their quality. Our added worry is narrower: case formulation is a skill built by struggling with a case, being wrong, and reworking it in supervision. Outsourcing that reasoning has costs that will not appear on any quarterly metric.
What about AI notetakers in the therapy room?
Ambient documentation tools are a different category from consumer chatbots, and APA’s advisory explicitly excludes provider-facing tools from its scope. That does not make them risk-free. Recording implies storage, vendor access, and a transcript that could be requested in a legal proceeding. Our position is that consent for any recording or transcription must be specific, genuinely optional, and revocable at any time, and that every client should be told plainly whether it is happening.
What can clients do about it?
Ask whether AI tools are used anywhere in your care, and exactly how. Ask whether your sessions are recorded or transcribed, by whom, and where the data is stored. Tell your therapist which tools you use, since APA recommends that disclosure in both directions. If an insurer or platform offers you an app instead of a clinician, ask what happens when the app is not enough. And if you have out-of-network benefits, look at them, because a narrow network is a pricing decision rather than a fact of nature.
Frequently asked questions
Do most therapists oppose AI? No. In APA’s 2026 survey, 54% were comfortable with some patients using chatbots and 40% were optimistic about their usefulness when a professional is unavailable. The objection is to substitution and to safety gaps, not to novelty.
Is my therapist allowed to use AI to write my notes? Practices vary, and policy is still catching up. You are entitled to ask what tools are in use, and to decline being recorded or transcribed.
Will AI make therapy cheaper for me? Not necessarily. Cheaper delivery has historically not translated into lower out-of-pocket costs for clients, and the RTI findings suggest the savings have been absorbed elsewhere in the system.
Sources
American Psychological Association. Psychologists say patients are turning to chatbots as mental health professionals, June 16, 2026. 2026 Chatbots and Mental Health Survey, 1,242 licensed psychologists, fielded April 9-26, 2026. apa.org/news/press/releases/2026/06/patients-chatbots-mental-health
American Psychological Association. Health advisory: Use of generative AI chatbots and wellness applications for mental health, November 2025. apa.org/topics/artificial-intelligence-machine-learning/health-advisory-chatbots-wellness-apps
American Psychological Association. APA alarmed by poor network access for mental health patients, April 17, 2024, citing RTI International, Behavioral Health Parity: Pervasive Disparities in Access to In-Network Care Continue. apa.org/news/press/releases/2024/04/poor-network-mental-health-patients
KFF State Health Facts. Mental Health Care Health Professional Shortage Areas (HPSAs), as of December 31, 2025. kff.org/other-health/state-indicator/mental-health-care-health-professional-shortage-areas-hpsas
Brooklyn Integrative Psychological Services is a group psychotherapy practice with offices in Greenpoint and Union Square, and virtual sessions across New York, Oregon, and Massachusetts. If you want to talk about what good care looks like for you, reach out for a consultation.