ChatGPT for Clinicians: Who Qualifies, What's Free, and Where the BAA Stops

TL;DR: free for the clinician, not a plan for the practice

If you're one clinician who wants cited answers and help with letters, sign up. It costs nothing, and it does more than the $20 Plus plan for this work. If you run a practice, it solves a smaller problem than the headline suggests. Every clinician gets a separate workspace, nobody administers them, and the staff who handle the most patient paperwork can't join.

I set up AI tools for small businesses. I'm not a clinician or your compliance attorney. This page is what OpenAI's own pages say, including the three places they don't quite line up.

Who qualifies, and where OpenAI's pages disagree

OpenAI describes eligibility in two different ways, and neither page mentions the other.

Pharmacists are on the first list and not named on the second. Psychologists are on the second and not the first. Therapists have the most riding on this, and I wouldn't guess. Try the verification. If it goes through, get the BAA in place before any client information goes in. If you're a counselor or social worker and it doesn't, the options are on Is ChatGPT HIPAA compliant?

To sign up you need a ChatGPT account, a valid NPI and a license the third-party verifier can confirm. It's US-only for now. It opens as a new workspace next to your existing ChatGPT account, and you move between them with the account switcher. That's worth remembering, because a chat in your personal workspace isn't protected by the BAA you signed in the other one.

What's free, and what isn't included

"Free" here is the launch wording. OpenAI's help article says it's free "at launch" and names no end date. ChatGPT for Teachers at least prints June 2028. A clinician who builds a routine around this should expect that the terms can change.

That last row catches people. OpenAI says not to put patient names, record numbers or other PHI into queries sent to the public healthcare sources, even with a BAA signed. Your conversations aren't used to train OpenAI's models. That's a separate promise from a BAA, and it doesn't make PHI safe to paste by itself.

How the BAA actually works: three limits the plan page doesn't spell out

You have to be authorized to sign it. OpenAI's wording is "Do not share PHI in ChatGPT for Clinicians unless a BAA is in place and you are authorized to sign a BAA for your account." A practice owner can usually sign for themselves. An employed physician at someone else's practice may have no authority to bind the employer, whose patients they are. Ask before you click, not after.

It stops at the person who signed. OpenAI says the product is "designed for individual use." A BAA that covers multiple users means ChatGPT for Healthcare. Front desk staff, billers and medical assistants don't have an NPI route in, and they're the people pasting appointment notes today.

One covered-features line needs a straight answer from OpenAI. Trusted clinical search is the main feature of this product. But OpenAI's list of functions covered under the BAA reads "Enable Trusted clinical search (Available ChatGPT for Healthcare only)." That may just describe an admin toggle, and I can't tell from the page. Until OpenAI says otherwise, I'd run evidence questions with no patient identifiers in them, which OpenAI itself notes covers many clinical tasks. Third-party GPTs and connected apps need the same caution, because they carry their own terms.

What this means for a small practice

Take an office with 3 clinicians and 5 staff, which is my example, not an OpenAI one.

So the free plan saves the 3 clinicians $60 a month against Plus and gets them something Plus can't offer. It does nothing for the 5 people at the desk. For them, the honest choices are no patient information at all, or a different covered tool. ChatGPT Business still works for staff on tasks with no PHI, such as policies, templates and website copy. If you're already on Microsoft 365 or Claude, compare Copilot and Claude, because each vendor draws the line in a different place. The wider map is on HIPAA-compliant AI.

The practical risk is the gap between the two groups. If the doctor has a covered workspace and the front desk has a personal account, the office can end up believing it's "on ChatGPT" safely when only one person is. Write down who may use what, with which information.

The micro-action for today

Ask every person in the office one question: which AI tool did you use last week, and on which account? Then sign up the clinicians who can sign for themselves, and tell everyone else in writing what stays out of ChatGPT.

If you'd rather have this planned and set up for the whole office, every price is on AI consulting for healthcare practices. The first step is a free 30-minute call, and if the answer is that you need counsel first, I'll say so.

Sources

• OpenAI Help, ChatGPT for Clinicians: eligibility, signup, features, BAA, unsupported features. Read November 2, 2026.
• ChatGPT for Clinicians plan page: the broader eligibility wording, HIPAA FAQ. Read November 2, 2026.
• OpenAI, Making ChatGPT free for clinicians: launch post. Read November 2, 2026.
• OpenAI Help, HIPAA eligible products and functionality: eligible products, covered and uncovered features. Read November 2, 2026.
• ChatGPT for Healthcare: comparison table and pricing statement. Read November 2, 2026.
• ChatGPT Business pricing: seat prices. Read November 2, 2026. The office table is my arithmetic. This is a technology explainer, not legal advice.