The AI Use Plan in R156-60e: What Utah Now Requires in Every Supervision Contract

Quick answer: Under R156-60e-307.1(2)(c)(iii), every clinical supervision contract executed under Utah's new rule must include a written, detailed plan for how artificial intelligence will be managed in practice. The plan has to address four areas — supervision, administration, note and report writing, and best practices — and it has to be signed by both the supervisor and the supervisee and kept in the supervisee's file.

This is the piece of R156-60e that's genuinely new, and it's the one most existing trainings are barely touching. So let's give it the room it deserves.

Why this section exists at all

Every other part of R156-60e has a long history. Supervision contracts, hours logs, ethics codes, reporting timelines — those have lived in Utah's rules in some form for years. The April 2026 rule cleaned them up and consolidated them, but they weren't new ideas.

The AI use plan is new. It showed up because the thing it governs showed up. Two years ago, the question of whether an AI tool drafted a progress note wasn't on anyone's radar. Now a newly licensed clinician can run a session through an AI scribe, get a structured note back in under a minute, and sign it before she leaves the building. That's a real shift in how clinical work happens — and the rule is responding to it the way good supervision always responds to a new capability: by asking you to be deliberate about it instead of letting it happen by default.

So the rule doesn't tell you whether to allow AI. It tells you that you and your supervisee have to decide, together, on purpose, and write it down.

The four domains the plan must cover

R156-60e-307.1(2)(c)(iii) names four specific areas your AI use plan has to address. Here's what each one is really asking.

(A) Supervision. How will AI tools be used — or not used — in the supervision relationship itself? In your prep for sessions? If you summarize a supervisee's case notes with an AI tool before a meeting, that belongs here. If you've decided AI stays out of the supervision room entirely, that belongs here too. Either answer is fine. The rule wants the answer named.

(B) Administration. The non-clinical work — scheduling, drafting routine communications, organizing documentation. This is usually the lowest-risk, highest-relief category, and it's where most practices land their first comfortable "yes."

(C) Note and report writing. This is the heart of it. Clinical documentation, progress notes, treatment plans, assessments. Are AI scribes permitted in your supervisee's clinical work? Under what conditions? Who reviews the AI's output before it goes into the chart? What gets disclosed to the client in informed consent? This is the domain with the most clinical and compliance weight, and it deserves the most specific language.

(D) Best practices. The general practices both of you will follow as the tools keep changing. This is the part that keeps the plan from going stale the moment a new tool launches.

The questions a real plan answers

A compliant AI use plan isn't a checkbox that says "we will use AI responsibly." It's a set of practice decisions, made in advance, in writing. The questions worth working through together:

• Are AI scribes permitted in the supervisee's clinical work, and if so, under exactly what conditions?
• Who reviews AI-generated output before it becomes part of the clinical record?
• What gets disclosed to clients about AI use, and where does that disclosure live?
• What HIPAA Business Associate Agreement is in place for each AI tool that touches client information?
• What can a supervisee safely put into a consumer-grade chatbot — and what is strictly off-limits because no BAA covers it?

That last pair is where the most expensive mistakes hide. A BAA — Business Associate Agreement — is the contract that makes a vendor legally responsible for protecting the health information you share with it. A purpose-built clinical AI tool with a signed BAA is one thing. A free, general-purpose chatbot with no BAA is another entirely, and protected health information should never go near it. A good AI use plan draws that line in language a supervisee can actually follow on a Tuesday afternoon.

What you do with it once it's written

Three steps, and they're not optional. You write the plan. You and your supervisee both sign it. You keep it in the supervisee's file alongside the rest of the supervision contract.

Most supervisors already have an instinct about all of this. You probably have a gut sense of what you'd want a brand-new clinician doing or not doing with an AI tool. The rule's only real demand is that you move that instinct out of your head and onto a page that both of you have read and agreed to. That's not bureaucracy for its own sake. It's the same thing supervision has always done — turn a seasoned clinician's judgment into something a newer clinician can hold onto.

This is learnable, and it's worth learning

Here's the encouraging part. The AI use plan looks intimidating because it's new, not because it's hard. Once you've worked through the four domains once — with good template language and the conditional logic for when a BAA is required — you have a plan you can adapt for every supervisee after that. The first one takes real thought. The tenth takes twenty minutes.

And there's a quiet upside most supervisors don't expect. Working through the AI use plan with a supervisee turns out to be one of the richest supervision conversations you'll have all year. It surfaces how they actually think about client privacy, about shortcuts, about what they're tempted to outsource and what they'd never hand over. That's gold for a supervisor. The rule asked for a compliance document. What you get is a window into your supervisee's clinical judgment.

Frequently asked questions

Does R156-60e require an AI use plan? Yes. R156-60e-307.1(2)(c)(iii) requires every supervision contract under the new rule to include a detailed written plan for how AI will be managed in practice.

What must the AI use plan cover? Four domains: (A) supervision, (B) administration, (C) note and report writing, and (D) best practices.

Can supervisees use AI scribes under R156-60e? The rule doesn't ban or require it. It requires the supervisor and supervisee to decide whether AI scribes are permitted, under what conditions, and who reviews the output — and to put that decision in the signed plan.

Does the AI use plan need to be signed? Yes. Both the supervisor and supervisee sign it, and it's kept in the supervisee's file with the rest of the supervision contract.

What is a BAA and why does it matter for the AI use plan? A Business Associate Agreement (BAA) is the contract that makes an AI vendor legally responsible for protecting any health information it handles. Protected health information should only go into AI tools covered by a signed BAA — which is exactly the kind of line a good AI use plan draws.

The Integration Institute · Practices That Work. Work That Matters. This post is a working reference and is not legal advice. Citations are accurate as of the final adopted R156-60e (effective May 26, 2026).

Want the actual template? The Integration Institute's supervisor training includes a full AI use plan template — pre-drafted language for all four required domains, plus the HIPAA and BAA logic — and you leave having drafted your own version, ready to use with your next supervisee.

See the training → https://www.theintegrationinstitute.com/supervision-training