Most coverage of AI mental health tools lands in one of two places: breathless enthusiasm about democratizing access to care, or hand-wringing about chatbots replacing therapists. Both framings miss what’s actually happening in therapy rooms right now. People aren’t choosing between AI and a therapist. They’re using both, often without telling anyone, and the gap between those two experiences is becoming a clinical problem therapists are scrambling to address.

A landmark APA survey released in June 2026, polling more than 1,200 licensed psychologists actively seeing patients, found that 77% of psychologists now have patients who openly discuss using AI chatbots for mental health support. Openly. Which implies a larger group who don’t bring it up at all. Thirty-five percent of those psychologists say their patients are treating AI as an additional mental health professional, and 39% have had patients use AI to self-diagnose. These aren’t edge cases. This is standard practice now, and the clinical field is trying to catch up.

The Problem Isn’t That People Are Using AI. It’s How.

There’s a reasonable version of AI in mental health support. Practicing a cognitive behavioral therapy thought record between sessions, tracking mood patterns, having something to talk to at 2 a.m. when your therapist isn’t available and the distress is manageable. These are narrow, defined tasks, and experts consistently note that chatbots can be useful in exactly those contexts.

The problem is the gap between that reasonable version and what’s actually happening. Thirty-nine percent of psychologists surveyed have seen patients use AI to self-diagnose. That’s not mood tracking. That’s a patient arriving at a clinical conclusion about themselves based on a conversation with a system that isn’t licensed, can’t observe them over time, doesn’t have access to their history, and was almost certainly not designed for diagnosis in the first place. Some of those patients then bring that self-diagnosis into session as a fixed belief. Some don’t come to session at all, because the chatbot told them something that felt sufficient.

The AI isn’t malicious. It’s just not built for this. The distinction matters because it changes what the actual risk is. It’s not that the chatbot is lying. It’s that it’s optimized to be helpful and responsive in ways that can feel like clinical authority without carrying any of the accountability or the rigor.

What Therapists Are Actually Worried About

Recommended The CBT Workbook for Mental Health

Evidence-based CBT exercises by clinical psychologists to transform negative thoughts and manage well-being.

Check Price →
FindingPercentageSource
Psychologists with patients openly discussing AI chatbot use77%APA survey (June 2026)
Psychologists reporting patients treat AI as additional mental health professional35%APA survey (June 2026)
Psychologists reporting patients use AI for self-diagnosis39%APA survey (June 2026)
Psychologists expressing concern about patient chatbot engagement94%APA survey (June 2026)
Psychiatrists supporting age restrictions on AI mental health chatbots78%APA survey (April 2026)
Psychiatrists wanting FDA clearance required for AI mental health apps77%APA survey (April 2026)

Helpful resource: The Mindfulness and Acceptance Workbook for Anxiety is a top-rated option for this. (As an Amazon Associate this site earns from qualifying purchases.)

Ninety-four percent of psychologists surveyed expressed concern about some of their patients engaging with chatbots, specifically emphasizing the risk of relying on AI as a sole mental health resource. That’s not a professional guild protecting turf. That number is high enough to take seriously on its own terms.

The concerns cluster around a few specific issues. First, therapeutic alliance, which is the research-backed relationship between therapist and client that predicts treatment outcomes better than almost any other variable. When a patient is getting emotionally significant responses from an AI between sessions, and especially when they’re not disclosing that, it can subtly compete with the therapeutic relationship without either party fully understanding why. Second, crisis recognition. A licensed clinician is trained to notice when something has escalated and to act on it. A chatbot, even a well-designed one, doesn’t carry that responsibility the same way, and the consequences of missing a crisis signal are not equivalent.

Third, and maybe most practically, there’s the self-diagnosis problem again. A patient who’s convinced they have a particular diagnosis based on an AI conversation may resist assessment, push for specific treatments, or feel dismissed when a clinician offers a different picture. That’s not hypothetical. Therapists are dealing with it now.

The Psychiatry Field Wants Guardrails, Not a Ban

A separate APA survey of psychiatrists, published in April 2026, adds another dimension. Seventy-eight percent of psychiatrists support age restrictions on AI mental health chatbots, and 77% want FDA clearance required for AI mental health apps. That’s a strong majority in favor of regulatory structure, not prohibition.

This is an important distinction. The clinical field isn’t saying AI has no place in mental health support. It’s saying the current landscape, where any app can market itself as a mental health tool without clinical validation, is genuinely dangerous. FDA clearance for medical devices requires evidence that something does what it claims and doesn’t cause harm it hasn’t disclosed. Most mental health apps on the market right now meet neither bar. Psychiatrists are watching patients make treatment decisions based on tools that have never been tested the way any medication or clinical intervention would be.

The argument for age restrictions is particularly pointed given what we know about adolescent development and identity formation. A teenager using an AI as a primary emotional confidant is a different clinical picture than an adult using one as a supplement to weekly therapy.

How to Actually Have This Conversation with Your Therapist

If you’re using AI for any kind of mental health support and you haven’t mentioned it to your therapist, that conversation is worth having. Not because your therapist will object, but because it’s clinical information that’s relevant to your care. The APA has published guidance specifically on discussing AI use in therapy, which suggests the profession sees this as a normal part of intake and ongoing sessions now, not a confession.

Your therapist isn’t going to take your phone. What they can do is help you think clearly about what you’re getting from the AI interaction and whether it’s filling a gap that therapy itself should be addressing. Maybe you’re using a chatbot late at night because you feel like you don’t have enough session time. Maybe you’re using it because there’s something you’re not ready to say to a human yet. Both of those are things a therapist can work with.

What’s worth monitoring in yourself: if you’re making decisions about your mental health, changing how you think about your diagnosis, or postponing professional care because of something an AI told you, that’s the point where the tool has stopped being a supplement and started being a replacement. That distinction is exactly what 94% of surveyed psychologists are worried about, and it’s a reasonable thing to worry about.

The bottom line is that AI isn’t going away as part of people’s mental health lives. Therapists know this, most are trying to engage with it thoughtfully rather than dismiss it, and the clinical conversation is shifting in real time. But the gap between what these tools are built for and how they’re being used is real, and navigating it thoughtfully, with a licensed professional as the anchor, is still the standard of care. If you don’t currently have a therapist and you’ve been leaning on AI to fill that role, that’s the clearest signal in this whole conversation: it’s time to find one.

Sources

This article is for general informational purposes only and does not constitute mental health, medical, or clinical advice. If you are in crisis or experiencing a mental health emergency, please contact the 988 Suicide and Crisis Lifeline (call or text 988) or go to your nearest emergency room. Always consult a licensed mental health professional for care specific to your needs.

Disclosure: As an Amazon Associate, we earn a small commission from qualifying purchases at no extra cost to you. We only recommend products that genuinely support the topics covered in this article.