How to Talk to Clients About AI in Your Practice

Quick answer

Tell clients about AI tools in your informed consent form, explain what data the tools process and how you protect it, and give clients a clear opportunity to ask questions or opt out. Transparency satisfies your ethical and HIPAA obligations and, more practically, protects the therapeutic relationship before questions come up.

The first time a client asks "wait, is a computer writing my notes?" can catch you off guard if you haven't prepared a clear answer. More practices adopt AI documentation tools every year, and the disclosure conversation is becoming as routine as explaining confidentiality limits. This guide gives you practical language and a simple framework to handle it well, whether you're using AI for session notes, client intake, or both.

Why This Conversation Matters

Clients come to therapy trusting that their most sensitive information stays handled carefully. AI tools introduce a new processing layer in that chain, and clients deserve to understand it. Even if your tool uses strong encryption and operates under a Business Associate Agreement, a client who discovers AI involvement without prior notice may feel their trust was violated, and they would not be wrong to feel that way.

Disclosure also satisfies real professional obligations. The APA Ethics Code (Standard 4.02) requires informed consent that explains the nature of services. NASW standards call for transparency about technology use in service delivery. Several state licensing boards have issued specific guidance on AI in clinical practice, and more follow regularly. The ethical floor is rising.

Beyond compliance, clients who understand your workflow are less likely to feel alarmed mid-treatment. A brief, matter-of-fact explanation during intake normalizes the technology and protects the alliance you've worked to build.

What You Are Required to Disclose

HIPAA and AI documentation tools

HIPAA requires that clients receive a Notice of Privacy Practices explaining how their protected health information is used and shared. When an AI tool processes PHI, even temporarily, that vendor almost certainly qualifies as a business associate under HIPAA. You need a signed Business Associate Agreement with that vendor, and your privacy notice should reflect that AI tools participate in your documentation workflow.

You don't need to write a technical specification into your consent form. A clear, plain-language statement suffices: that AI-assisted tools support clinical documentation, that those tools operate under a BAA, and that PHI is handled in accordance with your privacy policy.

State and ethics board requirements

Rules vary significantly by state. Some have issued guidance requiring explicit written consent for AI use in clinical contexts. Others haven't addressed it yet, but general informed consent and transparency standards still apply. Check your state licensing board's current guidance and your malpractice carrier's recommendations before finalizing your policy.

When in doubt, disclose more rather than less. No client has ever ended therapy because their therapist was straightforward about their documentation tools.

How to Update Your Informed Consent

Your existing informed consent form is the right place for AI disclosure. You don't need a standalone document. Keep the language brief and free of jargon.

Sample language you can adapt

"This practice uses AI-assisted tools to support clinical documentation. These tools process session-related information to generate draft notes, which I review, edit, and approve before they become part of your record. All tools are HIPAA-compliant and operate under a Business Associate Agreement. Your information never trains AI models. You may request manual note-taking at any time."

Adjust the specifics to match the tools you actually use. If you use PsyFiGPT for SOAP or DAP note generation, you can name it directly. If you use PsyFi Assist for intake, note that the initial intake process involves an AI assistant.

Some practices add a checkbox:

"I have read and understood the above disclosure regarding AI-assisted documentation. [ ] Yes [ ] No — I would like to discuss this before proceeding."

This creates a documented record and opens the door for questions before the first session begins.

Introducing AI During an Intake Session

Your consent form covers the legal and ethical baseline. The verbal conversation during intake builds the trust behind it.

A brief script for in-session introduction:

"Before we start, I want to mention that I use an AI tool to help with the documentation side of my practice. It helps me write accurate, organized notes after our sessions, and I review and edit everything before it goes into your record. Your information stays private and I don't share it with third parties. Any questions about that?"

If you use an AI intake tool like PsyFi Assist, introduce it at the start of the intake process:

"You'll see a brief questionnaire in your client portal. An AI assistant helps gather that information so I can come to our first session prepared. Your answers are confidential and go directly into your file for my review."

Use the same matter-of-fact tone you'd use to explain any other administrative step. Treating this as a casual part of onboarding signals to clients that it's routine, not something to be concerned about.

When Clients Push Back

Some clients will have concerns. Most of these are manageable with honest, direct responses.

"I don't want AI involved in my care." Acknowledge the preference without arguing. Explain what the tool actually does (it writes draft notes; it doesn't make clinical decisions or communicate with anyone). If the client still objects, clarify what your opt-out options are. Most clinicians can accommodate manual documentation, though it may affect scheduling if session time gets used for note-writing.

"Who sees my data?" Walk through the data path: session context goes into the tool, a draft note comes out, you review and approve it, and it enters your EHR. The AI vendor stores nothing beyond what the BAA permits. No third parties receive the data.

"Could this be hacked?" Purpose-built clinical AI tools use encryption at rest and in transit. Ask your vendor for specifics and share them if a client asks. Being able to answer this question concretely builds more confidence than a general reassurance.

"Is the AI listening to my sessions?" Clarify how the tool actually works. Most AI documentation tools, including PsyFiGPT, work from structured prompts or brief session summaries that you enter after the session, not from live audio recording. If yours does use audio, say so clearly and explain the privacy protections in place.

The Clinical Relationship Stays Yours

One clarification worth making in almost every disclosure conversation: AI handles the paperwork. It does not shape your clinical judgment, interpret what clients say, or make treatment decisions.

Your assessment, your diagnostic reasoning, your decision about what to include or exclude from a note, and your therapeutic relationship with the client are entirely yours. The AI produces a draft structure; you apply the clinical thinking that makes the note accurate and useful.

Framing it this way addresses the underlying concern many clients have, which isn't really about HIPAA. They want to know their therapist is actually present in the room and actually thinking about them. Confirming that AI handles documentation logistics, not clinical work, resolves that concern directly.

Frequently Asked Questions

Do I have to tell clients I use AI for notes? Yes, in most cases. Ethical guidelines from APA and NASW recommend disclosing AI use that affects client data. HIPAA also requires transparency about how PHI is processed, which includes AI documentation tools.

What should my informed consent say about AI? Include the name or type of tool, what data it processes, how it is protected (encryption, HIPAA compliance), whether output is reviewed by you before use, and how clients can opt out.

Can clients refuse AI documentation? Yes. Clients can request manual note-taking. Whether that changes your ability to continue seeing them depends on your practice policies, which should be stated clearly in your consent form.

Is AI-generated documentation HIPAA compliant? It can be, provided the tool operates under a Business Associate Agreement with your practice. Tools like PsyFiGPT are built with HIPAA alignment, including PHI de-identification and encrypted storage.

What if a client asks to see the AI output? Clients have the right to access their records under HIPAA. If AI-generated content becomes part of the clinical record after your review and approval, it follows the same access rules as any other note.

How do I explain an AI intake tool to a new client? Keep it simple. Tell them an AI assistant gathers initial information to help you prepare for the first session, that their data is protected, and that you review everything before it enters their file.

What is a BAA and do I need one for AI tools? A Business Associate Agreement is a contract that holds a vendor legally accountable for protecting PHI under HIPAA. If any AI tool processes PHI, a BAA is required. Reputable platforms like PsyFiGPT include BAA provisions as part of their service agreement.

Should I use the same AI disclosure for telehealth and in-person clients? The core disclosure should be the same for both. For telehealth, you may want to note any differences in how data transmits during remote sessions, since those involve additional technical channels.

Transparent disclosure about AI tools is a straightforward part of modern clinical practice. Handled well during intake, it takes less than two minutes and rarely generates concern. Handled poorly, or not at all, it creates problems that are much harder to address mid-treatment.

Questions about how PsyFiGPT handles client data and BAA requirements? Contact us or review the privacy documentation at psyfigpt.com.