Billing

Private Practice

Groups

Can Illinois Therapists Use AI for Notes and Billing? What HB1806 Actually Allows.

By Dax Earl • June 1, 2026

Last updated: June 1, 2026.

Illinois is not banning useful automation from therapy practices. It is drawing a line between AI that helps run the practice and AI that acts like the therapist.

That distinction matters for clinicians who are already using, evaluating, or being pitched AI scribes, EHR automations, billing helpers, intake summaries, and client messaging workflows.

Practice-operations takeaway: use AI for the back office, use AI documentation only with clear consent and therapist review, and do not let AI talk to clients like a therapist.

This article is general operational education, not legal advice. Illinois practices should review their specific workflows with counsel or compliance support.

Sections

TL;DR

HB1806 became Illinois Public Act 104-0054 and took effect on August 1, 2025. Read the public act.
The law does not ban every AI tool from a therapy practice. It allows AI for administrative support and certain supplementary support when the licensed professional remains responsible.
AI cannot provide therapy, make independent therapeutic decisions, directly interact with clients in therapeutic communication, generate treatment recommendations or treatment plans without licensed-professional review and approval, or detect emotions or mental states.
Billing, insurance claims, scheduling, reminders, and general logistics are treated differently from therapeutic communication.
If an AI tool records or transcribes a therapeutic session, Illinois requires written notice, a specific purpose, and client consent.

What Changed?

Illinois created the Wellness and Oversight for Psychological Resources Act, a law focused on AI use in therapy and psychotherapy services. The public act says its purpose is to protect people seeking therapy or psychotherapy by making sure those services are delivered by qualified, licensed, or certified professionals, not unlicensed providers or unregulated AI systems. Read the statute.

IDFPR described the law as limiting AI in therapy and psychotherapy services while still allowing AI for administrative and supplementary support services for licensed behavioral health professionals. Read IDFPR's announcement.

That is the important distinction. The law does not say, "no AI in a therapy practice." It says AI may help with administrative or supplementary support, but therapy still has to be conducted by a licensed professional.

That distinction is the whole article.

The Key Question: Is AI Helping the Practice, or Acting Like the Therapist?

For therapists, the easiest way to understand the law is to ask:

Is the AI helping run the practice, or is the AI doing therapy?

If the AI is helping with scheduling, claims, reminders, billing logistics, payer follow-up, or internal task organization, that is much closer to administrative support.

If the AI is helping draft a note for the therapist to review, that may be supplementary support.

If the AI is responding to a client's emotional distress, suggesting coping strategies, adjusting treatment goals, recommending interventions, or analyzing the client's mental state, that is the danger zone.

Illinois defines therapeutic communication broadly. It includes interactions intended to diagnose, treat, or address mental, emotional, or behavioral health concerns. The statute includes examples like reflecting a client's thoughts or emotions, offering therapeutic strategies, providing emotional support, collaborating on treatment goals, and offering behavioral feedback. See the statutory definition.

That means a tool does not have to call itself therapy to create risk. The label matters less than the behavior.

What AI Uses Are Clearly Safer?

The safest category is administrative support. Illinois expressly includes appointment scheduling and reminders, billing and insurance claims, and general logistics communications that do not include therapeutic advice.

For a therapy practice, the boring AI use cases are often the cleanest ones:

appointment reminders
scheduling workflows
claim status summaries
billing queue organization
denial categorization
eligibility task support
general logistics messages
internal payer follow-up summaries
insurance workflow task routing
credentialing checklist support
CAQH or roster maintenance reminders

That is especially important for Bomi customers. Bomi's work is the insurance backend: benefit checks before sessions, claims after sessions, denial follow-up when things get stuck, and credential maintenance as practices grow. See how Bomi handles billing.

A tool that helps organize claim follow-up is not the same as a tool that tells a client how to process grief. A tool that helps summarize eligibility is not the same as a tool that suggests a diagnosis. A tool that flags a stale claim is not the same as a tool that changes a treatment plan.

The cleanest AI boundary: AI can help make the back office less chaotic. It should not become the clinician.

What About Billing and Insurance Tools?

This is where the law is surprisingly practical. Illinois explicitly includes processing billing and insurance claims inside administrative support.

That matters because many therapists hear "AI therapy law" and immediately wonder whether every automation tool in their practice is now risky. The answer is no.

A billing workflow can still use technology. A practice can still automate repetitive administrative work. A claims process can still be supported by software. A group practice can still use reporting tools, payer dashboards, denial queues, and revenue workflows.

But a billing or admin tool should stay in the billing or admin lane. It can help answer:

Did the claim submit?
Did the payer reject it?
Is the client's coverage active?
Is the deductible remaining?
Which claims are aging?
Which payer is underpaying?
Which clinician has credentialing issues?
Which client has an outstanding balance?

It should not answer:

What intervention should this client receive?
Does this client need a higher level of care?
What diagnosis best explains this presentation?
How should the client manage emotional distress?
Should the client change therapeutic goals?
What does the client's tone suggest about their mental state?

Good billing support should make insurance quieter. It should not create a second clinical system hiding inside the admin workflow.

What About AI Notes and Scribes?

AI scribes and note tools are not automatically banned. But they are not the same as pure billing or scheduling tools.

Illinois defines supplementary support to include preparing and maintaining client records, including therapy notes. It also includes analyzing anonymized data to track progress or identify trends, subject to licensed-professional review, and identifying or organizing external resources or referrals.

So yes, an AI documentation tool may fit within supplementary support. But if the tool records or transcribes the therapeutic session, the consent rule kicks in.

The patient, or the patient's legally authorized representative, must be informed in writing that AI will be used and told the specific purpose of the AI tool. The patient must also provide consent.

Illinois defines consent as a clear, explicit affirmative act. It has to be express, freely given, informed, voluntary, specific, unambiguous, written, and revocable. Electronic written consent can count. A broad terms-of-use agreement with AI language buried among unrelated information does not count.

That means a vague line in an intake packet is probably not the standard to aim for. A stronger consent workflow should make the AI use obvious.

Our practice may use an AI-assisted documentation tool to help prepare clinical notes from session audio or transcripts. The tool is used to support documentation only. Your therapist reviews and remains responsible for the final clinical record. The tool is not used to provide therapy, make treatment decisions, communicate with you therapeutically, or generate treatment plans without clinician review and approval. You may revoke consent for this use.

That is not magic language. It is a starting point for the kind of clarity Illinois appears to require: what the tool does, why it is used, and who remains responsible.

Why This Matters for Claims, Audits, and Revenue

Clinical notes are not just clinical notes. They are also part of the reimbursement record.

A note supports the service billed. It helps explain medical necessity. It can matter in a payer audit. It can matter when a payer questions a code, denies a claim, requests records, or reviews treatment patterns.

That means AI documentation tools can create revenue risk even when the tool is not directly involved in claims submission.

If an AI scribe invents details, overstates interventions, mislabels symptoms, creates boilerplate medical-necessity language, or generates a note the therapist does not carefully review, the practice may end up with documentation that does not match the actual session.

That is a billing problem.

Not because the AI submitted the claim. Because the claim sits on top of the clinical record.

For Bomi customers, this distinction matters. Bomi can help with benefit checks, claims, denials, EOBs, balances, CAQH, attestations, rosters, and revenue reporting. But the licensed clinician still owns the clinical service, code selection, treatment plan, and note.

Insurance operations can be supported. Clinical judgment cannot be outsourced.

What AI Uses Are Prohibited?

Illinois is much stricter once AI crosses into clinical decision-making or therapeutic communication. A licensed professional may not allow AI to:

make independent therapeutic decisions
directly interact with clients in any form of therapeutic communication
generate therapeutic recommendations or treatment plans without review and approval by the licensed professional
detect emotions or mental states

This affects more than obvious AI therapist apps.

It may also affect tools like:

intake tools that suggest diagnoses
chatbots that offer emotional support
AI tools that recommend interventions
automated treatment-plan generators
AI-generated homework sent directly to clients
tools that classify client emotion, risk, or mental state
portal assistants that respond to distress therapeutically
between-session support bots that provide coping advice or reassurance

The riskiest pattern is direct-to-client AI that sounds warm, helpful, and therapeutic. That is exactly what many AI products are designed to do. In Illinois, that is also where the line gets dangerous.

A Practical Way to Classify AI Tools

For a therapy practice, the law becomes easier to operationalize if every AI tool goes into one of three buckets.

Green Zone: Administrative AI

This includes scheduling, reminders, billing, claims, eligibility, payment logistics, internal task routing, and general non-therapeutic communications.

The key test: the tool is not giving therapeutic advice, making clinical judgments, or communicating with the client about mental health treatment.

Yellow Zone: Documentation and Supplementary Support

This includes AI scribes, note drafting, session summaries, intake organization, referral organization, or anonymized progress trend analysis.

The key test: the therapist reviews the output, remains responsible, and obtains written consent when the tool records or transcribes a therapeutic session.

Red Zone: Therapeutic AI

This includes AI that provides emotional support, reflects feelings, suggests coping skills, recommends treatment, modifies goals, detects mental state, or interacts with the client as part of care.

The key test: if the client experiences the AI as part of therapy, slow down.

Or more bluntly: if the AI sounds like a therapist, treat it as a problem.

What Solo Therapists Should Do Now

Solo therapists should start with a simple inventory. You do not need a 40-page AI policy to take this seriously. But you do need to know what tools are touching the practice.

Ask:

Am I using an AI scribe?
Does it record sessions?
Does it transcribe sessions?
Do I have written client consent?
Does the tool draft notes?
Do I review every note before signing?
Does the tool generate treatment recommendations?
Does the tool send anything directly to clients?
Does the tool detect emotions, mood, sentiment, or mental state?
Does the vendor use my data to train models?
Does the vendor sign a BAA if PHI is involved?
Can I disable risky features?

The goal is not to become anti-technology. The goal is to keep the workflow honest. AI can reduce paperwork. It should not quietly rewrite the clinical relationship.

What Group Practices Should Do Now

Group practices have a bigger problem: tool sprawl. One clinician might use an AI scribe. Another might use an EHR beta feature. Another might paste intake summaries into a general AI tool. Another might enable a client messaging assistant. Another might use an AI tool for supervision notes, treatment planning, or documentation cleanup.

The owner may not know any of this is happening. That is where group practices get exposed.

For group practices, the right move is to create a basic AI governance workflow. Not a giant bureaucracy. Just a clear operating system.

Create an AI tool inventory. List every AI tool used by the practice, including EHR features, transcription tools, note tools, intake tools, billing tools, marketing tools, and client messaging tools.
Classify each tool. Put each tool into administrative, supplementary, or therapeutic/prohibited. If a tool does more than one thing, classify each feature separately.
Standardize consent. If any clinician uses AI to record or transcribe sessions, the practice should have a consistent written consent workflow.
Review vendor settings. Some vendors bundle note drafting, treatment-plan suggestions, client messaging, and emotion detection together. Know what is enabled.
Train clinicians on the boundary. AI can help with admin. AI can help draft documentation with consent and review. AI cannot become the therapist.
Keep billing and clinical review connected. If AI is helping with notes, someone still needs to confirm that documentation supports the service billed.

How This Connects to Bomi's Customers

Most Bomi customers are not asking, "Can I replace myself with AI?" They are asking a more practical question: "How do I run a sustainable insurance-based practice without drowning in admin?"

That is the right question.

Insurance-based therapy practices already carry too much operational weight:

benefit checks
eligibility
claims
EOBs
denials
stale claims
payment plans
client balances
CAQH
attestations
payer rosters
credentialing maintenance
revenue reporting

For group practices, the work multiplies across clinicians, payers, locations, and effective dates. Bomi's group-practice support is built around that complexity: provider onboarding, rosters, claims by clinician and payer, denial and A/R reporting, revenue attribution, and credentialing maintenance. See Bomi for groups.

Illinois' AI law reinforces an important operating principle: the practice can get help with the insurance machine. The therapist still owns the therapy.

That is where Bomi fits. Bomi is not trying to be the clinician. Bomi is trying to make the insurance side less chaotic so clinicians can keep their attention where it belongs.

What This Law Is Not

This is not a ban on billing software.

This is not a ban on claim automation.

This is not a ban on appointment reminders.

This is not a ban on AI-assisted documentation.

This is not a requirement that therapists handwrite every note.

This is not a reason to panic-delete every tool.

It is also not permission to let a vendor define the rules for you. If a vendor says, "our AI is just support," the practice still needs to ask what the tool actually does.

Does it record sessions?
Does it transcribe sessions?
Does it draft notes?
Does it recommend treatment?
Does it message clients?
Does it offer emotional support?
Does it detect mental states?
Does it require consent?
Does the therapist review and approve the output?

The workflow matters more than the marketing page. Baker Donelson's analysis makes the same broad point: Illinois' law creates strict conditions for how licensed professionals may incorporate AI into behavioral health care, not just a branding rule for AI therapy apps. Read Baker Donelson's analysis.

The biggest operational miss will probably be consent. A lot of practices will hear, "AI documentation is allowed," and stop there.

But for session recording or transcription, Illinois requires written notice and consent. The notice has to explain that AI will be used and the specific purpose of the tool. Consent also has to be specific, written, informed, voluntary, unambiguous, and revocable.

That has real workflow implications. Practices may need to update:

intake forms
consent packets
EHR templates
telehealth scripts
documentation policies
clinician onboarding
vendor review checklists
client FAQs
revocation workflows

This is especially important for groups. If one clinician uses an AI scribe and another does not, the consent process needs to reflect that. If the practice changes tools, the specific-purpose language may need to change too. If a client revokes consent, the clinician needs a workflow that does not break documentation.

Consent is not just a form. It is an operating process.

A Simple AI Policy Starting Point

A useful internal policy can be short. This is not a final policy for every practice, but it is a practical starting point:

AI may be used for administrative support such as scheduling, billing, insurance claims, eligibility workflows, logistics messages, and internal task organization.
AI may be used for documentation support only when the licensed clinician reviews and approves the output before it becomes part of the record.
If AI records or transcribes a session, the client must provide written, specific, revocable consent before use.
AI may not directly communicate with clients in a therapeutic way.
AI may not make independent therapeutic decisions.
AI may not generate treatment recommendations or treatment plans without licensed-professional review and approval.
AI may not detect emotions or mental states.
Any AI tool that touches PHI must go through vendor review before use.
Clinicians may not paste identifiable client information into unapproved AI tools.
The licensed clinician remains responsible for clinical judgment, documentation, diagnosis, treatment planning, and client communication.

What Providers Can Do Now

If you are an Illinois therapist or group practice owner, start with the tools already in your workflow.

Review Your AI Scribes

If a tool records or transcribes sessions, make sure your consent process is specific, written, and revocable.

Review Your EHR Settings

Some AI features may be embedded inside systems you already use. Check whether any AI note, summary, intake, message, or treatment-plan feature is enabled.

Review Client-Facing Tools

Be careful with chatbots, portal assistants, and between-session support tools. The more therapeutic the interaction feels, the more risk it creates.

Review Billing and Admin Tools

Billing, claims, scheduling, and logistics support are treated differently from therapy. Keep those tools in the administrative lane.

Review Vendor Language

If a vendor promises therapy-like support, 24/7 emotional support, personalized coping advice, mood detection, or automated treatment planning, slow down.

Review Group-Practice Consistency

Group owners should not assume every clinician is using the same tools in the same way.

Bottom Line

Illinois therapists can still use AI in their practices.

But the law makes the boundary clearer.

AI for billing, scheduling, claims, reminders, and general logistics is much safer because it fits into administrative support.

AI for notes and documentation may still be allowed, but if the tool records or transcribes sessions, therapists need written, specific, revocable consent and the licensed professional must stay responsible for the output.

AI that talks to clients like a therapist, makes independent therapeutic decisions, generates treatment recommendations without review, or detects emotions or mental states is where the law gets strict.

For Bomi customers, the lesson is not "avoid technology." The lesson is: use technology to make the back office cleaner, keep therapy in the hands of licensed clinicians, make consent clear, review vendors carefully, and do not let AI turn a billing shortcut into a clinical risk.

Want Bomi to Handle the Insurance Backend?

Bomi helps therapy practices with benefit checks, claims, EOBs, denials, stale claims, balances, CAQH, attestations, rosters, credentialing maintenance, and revenue reporting.
You keep the clinical relationship. Bomi helps keep the insurance side moving. Talk to Bomi.

FAQ

Can Illinois therapists use AI for billing?

Yes. Illinois Public Act 104-0054 includes processing billing and insurance claims as administrative support, which is treated differently from therapeutic communication.

Can Illinois therapists use AI scribes?

AI scribes are not automatically banned, but if the tool records or transcribes a therapeutic session, the practice needs written notice, a specific purpose, and client consent. The licensed therapist must remain responsible for the output.

Can AI chat directly with therapy clients in Illinois?

Not in therapeutic communication. The law prohibits AI from directly interacting with clients in therapeutic communication and from making independent therapeutic decisions.

Can AI generate treatment plans?

AI may not generate therapeutic recommendations or treatment plans without review and approval by a licensed professional.

What should therapy group practices do first?

Start with an AI tool inventory, classify each tool by feature, standardize consent for recording or transcription, review vendor settings, and train clinicians on the boundary between administrative support and therapeutic communication.

Sources

Growing a group practice?

Bomi helps coordinate insurance operations across providers, payers, rosters, claims, denials, and reporting.

See group support