The leaders of the organization that serves the country's state medical boards published a direct answer on Monday to a question state legislatures have started asking. Artificial intelligence, they wrote, should not be licensed to practice medicine the way a physician is.
The position carries weight because of who is stating it. Humayun J. Chaudhry, president and chief executive of the Federation of State Medical Boards, and Christy Valentine Theard, chair of the FSMB board of directors, wrote in STAT that generative AI should be governed inside existing professional accountability structures rather than treated as a separate practitioner.
FSMB is advisory rather than regulatory. It represents the state medical and osteopathic boards of the United States and its territories, the bodies statutorily charged with licensing physicians and protecting the public.
The Argument Against a Separate License
Chaudhry and Valentine Theard framed licensure as something other than a competence certificate. Medicine rests on a social contract, they wrote, in which society grants physicians unique privileges in exchange for extraordinary duties of competence, ethics and accountability. A license, in that framing, is not simply a permit to generate thoughtful and informed answers.
The distinction they draw is between licensing human clinicians and authorizing, registering or otherwise approving AI tools. Responsibility, in their view, should scale with how autonomously a system operates, and physicians should remain accountable for harm caused by inappropriate reliance on AI.
That framework is already reflected in nonbinding policy guidance the FSMB House of Delegates adopted in 2024 on the responsible incorporation of AI into clinical practice. The guidance addresses accountability for AI use across clinical settings, encourages continuing education focused on AI in health care, and keeps professional responsibility with the licensee.
Legislation and a Pilot Program Forced the Issue
The essay is a response to concrete state activity, not a hypothetical debate.
Bills introduced this year in Idaho and Iowa raised the possibility of creating a state licensing structure for artificial intelligence-augmented and autonomous service providers, sitting apart from the existing medical board system. Idaho's House Bill 945 would have created a new Board of Autonomous Medical Practice with its own licensure program and regulatory sandbox. Iowa's House Study Bill 766 proposed a broad licensing and oversight structure for the same category. Neither has been enacted. The authors argue that their introduction is the significant fact, because it demonstrates the proposal is no longer theoretical.
Utah moved further. In January, the state's Office of Artificial Intelligence Policy announced a 12-month agreement with the health technology startup Doctronic allowing an autonomous AI platform to participate in prescription renewals for patients with chronic conditions such as hypertension, diabetes and thyroid disease, inside the state's regulatory sandbox. The arrangement removes the requirement that a clinician sign off on each individual renewal, with uncertain cases escalated to human clinicians.
It did not go unchallenged. The Utah Medical Licensing Board called for the program to be suspended over patient safety concerns, which the FSMB leaders cite as evidence that medical boards need a voice whenever clinical decisions and patient safety are involved, even when an innovation moves lawfully through a state technology office.
FSMB has formed a Workgroup on the Regulation of AI in the Practice of Medicine to update its guidance, because the existing framework predates the current wave of systems that push AI toward autonomous action.
The Serious Counterargument
This is a live disagreement among credentialed people, not a settled question, and the opposing case has been published in peer-reviewed venues.
In JAMA Internal Medicine, Eric Bressman of the University of Pennsylvania and colleagues Carmel Shachar, Ariel D. Stern, and Ateev Mehrotra proposed licensing AI in much the way clinicians are licensed, arguing that concerns about hallucination and performance drift resemble late 19th-century worries about variable clinician training that licensure was built to address. Their framework combines practice standards with ongoing surveillance and continuing education, ideally overseen by a new federal digital licensing board, with the FDA retaining premarket review so developers would not face 50 separate state authorities.
A separate JAMA viewpoint by Alon Bergman, Robert Wachter and Ezekiel Emanuel proposed a four-part licensure framework for autonomous clinical AI: demonstrated competency, requiring a model to meet or exceed the median score of recent human test-takers on licensing examinations and then complete a supervised deployment phase analogous to residency, plus a defined scope of practice, ongoing monitoring and periodic renewal. Their argument rests in part on projected physician shortages that traditional workforce fixes cannot close quickly.
Both camps agree on the underlying risk. They disagree about whether adapting licensure or extending existing accountability is the safer path.
Implications for Patients Today
Nothing about your next appointment changes because of an opinion essay. But three practical points follow from where this debate currently stands.
First, if an AI system is involved in your care, a licensed clinician remains legally accountable for the decision under every state's current framework. That is the status quo the FSMB is defending, and it is what gives patients a path to recourse when something goes wrong.
Second, patients can ask. Whether an AI tool contributed to a diagnosis, a note, a triage decision or a prescription renewal is a reasonable question, and FSMB's guidance encourages physicians to be transparent about AI use in care.
Third, the distinction that matters is autonomy. A tool that drafts a note a physician reviews is a different regulatory object from a system that renews a prescription without per-case clinician review. The Utah pilot sits closer to the second category, which is why it attracted a medical board's attention.
Where the Decision Will Actually Be Made
State legislatures reconvene in January, and the Idaho and Iowa bills are unlikely to be the last of their kind. The FSMB workgroup's updated guidance will shape how individual boards respond, though the guidance is advisory and each board retains its own statutory authority.
The 2024 FSMB policy advised boards to examine how the practice of medicine is defined in their own jurisdictions, a question that becomes consequential as systems act with less supervision. That definitional work is the mechanism through which this debate produces actual rules.
For readers, the useful thing to watch is not the philosophical argument but the specific pilots. Prescription renewal, triage, and diagnostic suggestion are the tasks moving first, and they are the ones where a poorly handled routine decision can produce real harm.
Frequently Asked Questions
What did the Federation of State Medical Boards say? Its president and board chair wrote that AI is not ready to be independently licensed like a physician and should instead be governed within existing professional and institutional accountability structures.
Does FSMB have the power to decide this? No. FSMB is advisory. It serves the state and territorial medical boards, but each board retains its own statutory licensing authority, and state legislatures write the underlying law.
Has any state tried to license AI as a practitioner? Bills in Idaho and Iowa this year raised the possibility of a separate licensing structure for artificial intelligence augmented and autonomous service providers. Neither has been enacted.
What is the Utah pilot? Utah's Office of Artificial Intelligence Policy approved a 12-month program with the startup Doctronic allowing an autonomous AI platform to handle certain prescription renewals for chronic conditions without a clinician signing off on each request.
Who disagrees with the FSMB position? Researchers writing in JAMA Internal Medicine and JAMA have proposed adapting licensure for autonomous clinical AI, including competency testing, a supervised deployment phase, and a defined scope of practice.
Who is responsible if AI contributes to a medical error today? Under current state frameworks, the licensed clinician remains accountable. FSMB guidance addresses physician responsibility for AI use in care.
Can I ask whether AI was used in my care? Yes. Asking whether an AI tool contributed to a diagnosis, note or prescription decision is reasonable, and transparency about that use is what FSMB guidance encourages.
