Asking Clients About AI Chatbot Use

If you haven't asked your clients whether they use AI chatbots, some of them almost certainly do.

What the latest survey shows

The American Psychological Association's 2026 survey of more than 1,200 US psychologists found (APA, 2026):

77% have had patients who used AI for support, engagement or other reasons
39% had patients who used AI to self-diagnose
33% said patients used AI to help with their therapy or treatment
Among psychologists whose patients had a relationship with a chatbot, 68% said patients felt validated or supported, but 36% noticed patients becoming dependent on it, and 15% noticed distorted thinking or delusions linked to it
97% worry chatbots may reinforce negative behaviours or delusional beliefs

APA's position is clear: AI is not a safe or effective replacement for a qualified mental health professional. Its advice to the public is to verify information with a professional and tell their clinician if they use AI for support.

The survey only reflects what psychologists see in their own clients, so actual use may be higher, especially among teens.

Why it's worth asking

Chatbot use can be helpful, neutral or harmful, and you won't know which unless you ask. It can also tell you a lot: what your client is anxious about between sessions, what they're not yet saying to you, and where they look for reassurance.

How to bring it up

Make it routine and non-judgmental, the same way you might ask about sleep, alcohol or social media. For example:

"A lot of people use AI chatbots now. Have you used one for anything to do with how you're feeling?"
"What kinds of things do you talk to it about?"
"How do you usually feel after those conversations?"
"Has it ever told you something that worried you, or that didn't match what we've talked about here?"

Consider adding a question about AI use to your intake form, so the topic is normal from the first session.

What to listen for

Often fine or even helpful: - Using AI for reminders, journaling prompts or practising skills you've taught - Looking up general information and bringing questions to session

Worth exploring further: - Using a chatbot to self-diagnose, or to second-guess your treatment plan - Spending long periods talking to a chatbot, or preferring it to people - Describing the chatbot as a friend, partner or "therapist"

Act on promptly: - Signs the chatbot is reinforcing hopelessness, risky behaviour, or unusual or delusional beliefs - Any mention of self-harm or suicide in relation to chatbot conversations, which should go through your normal risk assessment process

Turning it into part of therapy

Review together. If the client is comfortable, look at what the chatbot said and discuss it.
Set shared guidelines. Agree what AI use is helpful for this client, for example practising a coping skill, and what isn't.
Redirect to people. If a chatbot is filling a gap in connection, that's a treatment goal in itself.
Document it as you would any other relevant behaviour.

For clinic owners

Give your team a short shared approach: a standard intake question, a few example prompts, and clear guidance on when chatbot use should trigger a risk review. It keeps practice consistent across clinicians.

This article is general information for clinicians and does not replace your professional judgment, ethics code or local guidance.