AI In Your Practice: What Clinicians and Supervisors Need To Know

Why This Episode Matters

AI is already showing up in therapy notes, treatment plans, client apps, agency systems, and conversations with clients. Clinicians and supervisors need this information because using a convenient tool without understanding its privacy, consent, and documentation risks can place client trust, protected health information, and a professional license at risk.

AI went from a someday conversation to a this week conversation very quickly.

I have been hanging back a little, watching where everything falls. I do not want to approach this topic with hype, fear, or a blanket statement that AI is either good or bad.

I want to give you tools.

Ethical AI use in clinical supervision begins with asking practical questions about confidentiality, informed consent, documentation, and professional responsibility. These are questions clinicians and supervisors can begin asking today.

Ethical AI Use in Clinical Supervision Starts With PHI

When clients speak in the therapy room, they are entrusting us with information that belongs to them.

I sometimes compare it to lending someone your car. You expect that person to keep it secure, use it responsibly, and prevent unauthorized people from accessing it. Protected health information, or PHI, requires that same level of care.

One myth I see all the time is that information becomes safe as soon as the therapist removes the client’s last name. That is not necessarily true. A first name, initials, occupation, family situation, location, or unusual event may still make the client identifiable.

The question is not simply whether you included a name. The question is whether the client, a family member, a neighbor, a colleague, or someone in the community could recognize the person from the details.

This becomes especially important when a clinician places information into an AI tool. Before entering anything, slow down and consider what the tool will receive. Does the information include a diagnosis, clinical history, treatment details, or appointment information? Has the technology been approved to handle protected information? Do you understand where the information goes after you submit it?

Removing a name does not automatically remove the privacy risk.

A BAA Is the Beginning of the Review

A business associate agreement, or BAA, matters. It is not the end of your due diligence.

I hear clinicians say, “The company says it is HIPAA compliant,” or, “They told me they delete the information.” Those statements should lead to more questions.

I do not know what deletion means in every digital system. Is the information removed from the active account? Is it retained in backups? Can employees or contractors access it? Is the data used to improve the software or train a future version of the tool?

The vendor should be able to explain how information is encrypted, where it is stored, who can access it, how long it is retained, and what happens if there is a breach. You should also understand how information is removed from backups and archived systems.

If the system generates treatment plans, assessments, interventions, or client reminders, ask how it makes those decisions. What clinical theory informs the recommendations? What information does the system prioritize? How does it respond when information is incomplete or inaccurate?

I see this all the time with new technology. The output looks polished, so we assume the process behind it must be sound.

A polished AI response is still a draft. The clinician remains responsible for reviewing the content and making the clinical decision.

For more guidance on protecting client information, review HIPAA compliance and documentation for counselors.

Explain AI Use During Informed Consent

AI use should not be buried in the fine print of a long consent packet.

If a device is listening during the session, explain it. If an AI tool will create a draft note, explain it. If information from the session will connect to an app that sends reminders to the client, explain that too.

Clients need to understand what technology is being used, what information the technology receives, and why the clinician is using it. They also need to know how their information is stored, whether they have a choice, and what happens if they do not consent.

This explanation should happen in plain language.

A clinician might say:

“I use an AI-supported tool to help create a draft of my session documentation. I review and edit the note before it becomes part of your record. I want to explain how the tool handles your information and give you an opportunity to ask questions.”

The exact wording will depend on the setting, the technology, and your state requirements. The observable behavior is what matters.

The clinician explains the process, invites questions, obtains agreement, and documents the conversation. A signature alone does not demonstrate that the client understood what was happening.

Put AI Expectations in the Supervision Contract

Supervisors may not work at the same location as their supervisees. A supervisee may work for a hospital, agency, group practice, or online platform that requires the use of particular technology.

Sometimes the supervisee will not have a choice about the system. That does not remove the need for supervision.

I would include language in the supervision contract requiring the supervisee to disclose any technology that records sessions, processes client information, generates documentation, suggests diagnoses, recommends interventions, or sends information to clients.

The contract should also clarify that the supervisor may review consent forms and technology policies from each practice site. Supervisees should report changes in employment settings, platforms, or documentation tools instead of waiting until a concern appears.

Here’s what I tell supervisors. Ask about technology regularly.

Find out whether the workplace has introduced a new AI tool. Ask what information the tool receives and how the client is informed. Review where consent is documented and how the supervisee checks the accuracy of generated content.

You should also discuss what happens when the AI recommendation conflicts with the supervisee’s clinical judgment. That is not only a technology question. It is a clinical reasoning and documentation question.

A strong supervision contract creates clear expectations before a problem occurs. Read more about building a clear and effective supervision contract.

Stay Curious When Clients Use AI

Clients are using AI too.

They may arrive with a symptom list, a suggested diagnosis, a request for accommodations, or specific language they believe they need for paperwork. The information may sound unusually clinical or may not match what the therapist has observed.

That moment can feel like whiplash.

The answer is not to shame the client or immediately dismiss the information. Saying, “That sounds like AI. Tell me what is really happening,” may close the conversation before the clinician understands what the client was trying to accomplish.

Take a stance of curiosity instead.

Ask what led the client to search for the information. Explore which parts felt accurate and which parts did not fit. Ask what the client hoped the answer would help them understand, then compare the generated language with what the client experiences in daily life.

These questions keep the therapeutic relationship open. They also help the clinician separate the client’s lived experience from the language produced by a tool.

AI can provide information. It cannot replace the empathy, observation, context, and critical thinking of a trained human clinician.

Three Actions You Can Take Today

You do not need a perfect AI policy before you begin strengthening your systems.

Start by reviewing your informed consent. Make sure it clearly explains whether AI is used and what happens to client information.

Next, review your supervision contract. Add expectations for disclosing, evaluating, and documenting technology use across every practice site.

Finally, review your vendors. Ask how they access, store, retain, process, and permanently delete protected information.

AI may eventually feel as ordinary as using the internet or turning on a light. We are not there yet.

Right now, clinicians and supervisors need to slow the process down enough to communicate clearly, document carefully, and protect client choice. That is the foundation of ethical AI use in clinical supervision.

Where This Blog Connects

• Harnessing The Power of AI: A Guide For Elevating Your Professional Goals
• The Benefits of AI in Marketing for Therapists
• Leveraging AI for Administrative Tasks
• AI for Therapist Notes: A Blessing or a Curse?

Want to learn more? Check out this month’s free resource from Kate Walker Training.

Want deeper support as you navigate AI, documentation, informed consent, and supervision systems? Inside the Step It Up Membership, I help therapists and supervisors think through the ethical and operational decisions that protect clients, strengthen documentation, and support sustainable practice growth.

Blog post by Kate Walker, Ph.D., LPC-S, LMFT-S
Creator of Texas’s first fully online 40-Hour LPC/LMFT Supervisor Training Course.
Kate is the founder of Kate Walker Training, where she helps counselors, supervisors, and practice owners build sustainable, ethical, and profitable businesses.
Ready to grow your supervision practice? Check out the Supervisor Training Courses.

This post is a written summary of a podcast episode hosted by Dr. Kate Walker. We use AI to help format the transcript and check for clarity and spelling, turning spoken content into an easier-to-read format.