Your state board has an opinion on AI charting now. Here is what it wants.

What actually changed in 2026

For years, AI in the exam room sat in a gray zone with no specific rules. That zone closed this year.

In 2026, more than 40 bills touching AI in healthcare were introduced across roughly 25 states. Texas is the clearest example, with SB 1188 and HB 149 carrying explicit AI disclosure requirements. The exact text varies state to state, but the through-line is consistent: legislators want patients to know when AI is involved in their care, and they want a human clinician clearly accountable for the result.

The practical reading for a working physician is that AI charting is no longer just a workflow choice. It is starting to carry disclosure and documentation obligations that your license touches. That is not a reason to stop using these tools. It is a reason to use them in a way that you can defend.

The rule the FSMB landed on

The most useful signal does not come from any single state. It comes from the Federation of State Medical Boards, which created a Workgroup on AI Regulation to give boards a shared framework.

Its guidance lands on a principle that should feel familiar: the physician is ultimately responsible. AI can assist, draft, and suggest, but it does not hold a license and it does not carry the accountability. You do. The FSMB pairs this with two expectations. First, document your rationale for accepting or rejecting AI output. Second, handle informed consent and disclosure of AI involvement where it applies (Holland and Knight summarized this guidance in May 2026).

This is the part worth internalizing, because it outlasts any individual bill. Whatever your specific state does, the board-level consensus is that the human stays responsible and shows the reasoning. Build to that standard and the state-by-state details become a formatting question, not a panic.

It is worth being clear about why boards landed here rather than on something stricter. They could have tried to certify tools or ban categories of use. Instead they put the weight on the licensed clinician, which is the lever they already control. A board cannot meaningfully regulate a software vendor in another state, but it can hold its own licensees to a standard of accountability and documentation. That is why the rule reads the way it does, and why it is unlikely to soften: it is the natural shape of how medical boards enforce anything.

The 2026 disclosure picture, in plain terms

You do not need to track all 40-plus bills. You need to know what category your state is moving into. Broadly, three patterns are emerging:

• Disclosure-required states, where you must tell the patient when AI is involved in their care or documentation. Texas, through SB 1188 and HB 149, sits here.
• States actively legislating, where bills are introduced but the final shape is still moving. Most of the roughly 25 states with 2026 activity fall in this group, so the safe move is to assume disclosure is coming.
• Quiet states, where no specific AI rule exists yet. Even here, the FSMB principle still applies through your board's general standards, because the physician is responsible regardless.

The honest takeaway is that the differences between these are smaller than they look. A physician who discloses AI involvement and documents their rationale is compliant in all three categories. That is why the habit, not the statute, is the thing to build.

The one habit the FSMB actually wants

Strip away the legislative detail and the FSMB is asking for one concrete behavior: when AI touches the chart, document why you accepted or rejected what it produced.

In practice that is a short, defensible line in your own workflow. If you take the AI-drafted note as written, your review and sign-off is the rationale. If you change it, a brief note on what you corrected and why shows the human judgment the board is looking for. The companion briefing on how doctors use AI for clinical notes safely treats this same review step as the core safety habit, and it maps cleanly onto what the FSMB now expects on the compliance side.

This is not extra paperwork for its own sake. It is the record that proves you stayed in the loop, which is exactly what protects your license if anyone ever asks.

The mistake to avoid is treating the AI draft as a finished product you merely approve. The boards are drawing a sharp line between a clinician who used a tool and a clinician who deferred to one. The first is practicing medicine with help. The second has handed judgment to software that cannot hold a license. Your rationale, even a one-line one, is what puts you on the right side of that line in the record. Skip it consistently and a reviewer has no way to tell which kind of physician you were on any given chart.

Where consent fits, and where it does not

Disclosure and consent are related but not identical, and it is worth keeping them straight. Disclosure means telling the patient AI is involved. Consent means getting their agreement to it, which some uses and some states require, especially for ambient scribes that record the visit. If the consent question is your live concern, the briefing on when AI scribes need patient consent covers that specific ground. This briefing stays on the broader board-and-statute picture.

The skill under the rules

Compliance frameworks will keep shifting through 2027 and beyond. Chasing each new bill is a losing game. The thing that holds steady is the underlying skill the FSMB is really pointing at: knowing where AI helps in your charting, where your judgment has to override it, and how to leave a clean record of that judgment.

A physician with that habit is already compliant with the rules that exist and the ones still being written. AI for Physician Notes builds that workflow and the documentation reflex that goes with it, and the two minute course quiz will point you to the right starting place for your practice.