Another state just added itself to the list of places with its own AI health care law. Another compliance memo just landed on somebody’s desk. It’s easy to file this under bureaucratic noise, something legal will sort out. It isn’t. A physician documents why she overrode an ambient scribe’s note. A nurse practitioner checks a disclosure banner before a generative AI-drafted message goes out to a patient. A physician assistant reviews an AI-flagged risk score before acting on it. None of that is paperwork. That’s what practicing medicine looks like now, one more small decision layered onto every patient encounter.
It adds up in a way that rarely shows up in the policy language itself. A clinician already carries the clinical judgment, the documentation burden, and the malpractice exposure. Now there’s a second layer running underneath all of it: which disclosure applies today, how much review counts as enough, and whether an override needs its own note. None of that was part of the job a few years ago. It is now, and it lands on the person seeing the patient, not the administrator who chose the tool.
Where the rules actually come from
There’s still no federal law that covers AI in health care comprehensively. The Food and Drug Administration (FDA) keeps clearing higher-risk AI devices through its existing device pathways, and the Centers for Medicare & Medicaid Services (CMS) has drawn some lines around AI in Medicare Advantage prior authorization. Past that, Washington has mostly stayed out of it. A December 2025 executive order told the Justice Department to go after state laws it sees as too burdensome and pushed for a lighter national standard, but nothing has actually passed.
Real physician voices, twice a week
Free, and one click to unsubscribe.
So states filled the space. By the end of 2025, more than thirty health care-specific AI laws existed across roughly twenty-one states, and 2026 has added even more. Several now require that a real clinician review any AI-involved coverage denial, not just an algorithm. Some require practitioners to review AI-generated clinical output before acting on it. Others require disclosure when generative AI writes something a patient reads, or restrict AI from delivering psychotherapy on its own. Each rule sounds reasonable in isolation. Stacked together, they become something a single clinician is expected to track in real time, between patients, without any system built to help them do it.
Where it actually lands
A clinician who practices in one state can learn that state’s rules and move on with some confidence. A clinician working across a health system, a telehealth network, or a national practice group is juggling something closer to fifty different rulebooks, and the switching happens invisibly, tool by tool, patient by patient. A documentation workflow built for one state can trip a completely different notice requirement the moment a patient logs in from across a state line. Nobody flags that in the exam room. The clinician just has to know.
And here’s what none of these laws actually solve. They tell a clinician what to disclose and when to review something. They don’t say who inside the health system owns the decision to deploy a given AI tool in the first place, who answers for it when it underperforms, or how that gets documented well enough to hold up to a malpractice carrier or a board inquiry later. All of that gets pushed down to the point of care by default because nobody higher up claimed it first. The clinician ends up the last line of accountability for a decision she didn’t make.
What would actually help
None of this argues against regulation. Patients deserve protection from opaque denial algorithms, chatbots posing as therapists, and tools that quietly underperform for populations they were never properly tested on. States stepped in because Washington was slow, and the harms were real. The problem isn’t that rules exist. It’s that nothing above the clinician is absorbing the weight of applying them.
Two things would change that. First, a national floor: clear federal rules on transparency, oversight for high-stakes decisions, and accountability, with room for states to handle what’s genuinely local instead of each writing its own disclosure language. Second, and this is the part health system leadership keeps skipping, someone inside the organization needs to actually own AI deployment decisions the way someone already owns credentialing. Until that ownership exists somewhere above the bedside, no amount of federal or state clarity will lift the burden off the clinician deciding, patient by patient, whether to trust the machine or override it.
Until that happens, clinicians are left with a narrower set of tools:
- Documenting the reasoning behind accepting or overriding an AI suggestion
- Knowing the disclosure and review rules in every state where a license is held or patients are seen
- Pushing vendors for real answers on how their tools handle state-specific requirements
- Pushing health systems, not just professional associations, to build the internal governance this moment actually requires
Medicine is hard enough without also practicing regulatory law between patients. AI can genuinely lighten the load and sharpen decisions. It will only do that once the accountability sits somewhere above the person at the bedside instead of defaulting there because nobody else claimed it.
Matt Hasan is an economist, AI strategist, and founder of aiRESULTS. He advises health systems, payers, and life sciences organizations on the strategic implications of artificial intelligence, digital transformation, and emerging technologies. Over a career spanning more than four decades, he has held leadership and advisory roles with organizations including AT&T, IBM, Deloitte, Capgemini, and Citigroup, and previously served on the faculty of New York University’s Stern School of Business.
Dr. Hasan’s work focuses on the intersection of technology, institutions, and human decision making, with particular emphasis on how AI is reshaping medicine, governance, leadership, and professional practice. He is the founder of The AI Humanist Movement and an advocate for Human-AI Synergy, a framework that views AI not merely as a tool, but as a cognitive partner capable of extending human capabilities.
His writing includes “A Profession at the AI Frontier: Medicine Must Reinvent Itself or Cede Ground,” published in Health Affairs Forefront. He shares updates on LinkedIn and Medium.
