A Theoretical Look at Whether Ignoring AI Could One Day Be Considered Malpractice

By the Editor-in-Chief, Concierge Medicine Today

An Editor’s note to readers on framing before you read further. No court in the United States has yet ruled that a physician’s failure to use an AI tool constitutes malpractice. This article does not claim otherwise. What follows is a first-principles examination of where malpractice law has historically gone when a new technology becomes widely adopted, what legal scholars and health law journals are already writing about AI and the standard of care, and what that pattern of thinking suggests physicians should be paying attention to now, not later. This is educational and informational content, not legal advice. Every physician should consult their own malpractice counsel and carrier about how these questions apply to their specific practice.

I. The Question Nobody Wants to Ask Out Loud

Every physician has heard the anxious version of the AI conversation. Will it replace me? Every physician has also heard the reassuring version. It won’t replace you, it will replace physicians who don’t use it.

That second line has been said publicly, on the record, by more than one person who leads organized medicine. Ohio State, Jefferson Health, and AMA leadership have all repeated some form of it over the past three years. Ohio State’s Karthik Sarma, ChatGPT’s early adopters in radiology, and both AMA presidents who have addressed the question since 2023 have converged on nearly identical language. Ehrenfeld: “AI will never replace physicians, but physicians who use AI will replace those who don’t.” Yehia, president of Jefferson Health: doctors who do not use AI should be open-minded, because his own system does not believe AI will replace all doctors but is already building around the assumption that non-adopters fall behind. [10][11][14]

That framing has mostly been treated as a career observation. A competitive one. Use it or get left behind in the marketplace.

First principles require a harder question underneath that one. If the standard of practice shifts far enough, does “falling behind” stop being a career problem and start being a legal one?

That is not a rhetorical flourish. It is exactly how malpractice law has behaved with every major diagnostic and treatment technology that came before this one.

II. How the Standard of Care Actually Moves

To understand where this could go, it helps to strip away the AI headlines and look at what a malpractice claim actually requires. Four elements, consistently, across nearly every U.S. jurisdiction: the physician owed the patient a duty of care, the physician breached that duty, the breach caused harm, and the patient suffered real damage as a result. [2]

The breach element is where the standard of care lives. And the standard of care is not fixed. It is defined, as the Suffolk University Journal of Health and Biomedical Law put it plainly in a January 2026 analysis, as the care that a reasonable, similarly situated physician would have provided under the circumstances. [2] That definition moves as the profession moves. It is not asking what was reasonable in 1990, or even in 2020. It is asking what is reasonable now, among your peers, given the tools that are actually available to you.

This is precisely why legal scholars writing about AI keep returning to the same word: doctrinal collapse. A 2026 analysis on algorithmic authority and the clinical standard of care describes it this way. AI technologies are beginning to blur boundaries between three legal categories that used to be distinct: malpractice negligence, which asks whether the individual physician deviated from standard care; vicarious liability, which asks what the hospital is responsible for; and product liability, which asks what the AI manufacturer is responsible for. When diagnostic reasoning is shared between a clinician and an algorithm, those clean lines start to disappear. [7]

The same analysis names a scenario worth sitting with directly. Physicians may eventually face liability exposure on both ends of the same decision: liable for incorrectly relying on a flawed AI recommendation, and equally liable for failing to use an available, highly accurate AI tool that was sitting right there. [7]

Neither side of that trap is theoretical anymore. It is being written about in legal literature today, before a single landmark case has forced the question.

III. Medicine Has Lived This Exact Story Before

This is not the first time a technology threatened to move the floor under the standard of care.

Ask any radiologist practicing today. Imaging, once read almost entirely by eye and instinct, is now shaped by decision-support software that most radiologists rely on daily without a second thought. The same litigators who now write about AI liability point out plainly that as AI becomes more accurate and more widely available, the legal argument shifts. Attorneys may eventually argue that a physician was negligent not for using AI, but for underutilizing an advanced tool that was already the accepted standard among peers. [8]

That is the quiet pattern underneath every technology shift in the history of the standard of care. The new tool arrives. It is optional, then common, then expected, then simply assumed. Nobody sues a physician in year one for not adopting a brand-new tool. Physicians get sued, eventually, for not adopting a tool that has become ordinary among their peers while they kept practicing as though it did not exist.

That is not an AI story specifically. That is a medicine story, repeating itself.

IV. What the Legal Research Already Shows

A 2021 study published through the Journal of Nuclear Medicine, authored by legal scholars at Georgetown University Law Center and ETH Zürich, examined how jurors actually respond when physicians follow or reject AI treatment recommendations. Their concern was the opposite of the one physicians usually raise. Legal scholars had warned that following a nonstandard AI recommendation could increase a physician’s liability risk, which would quietly discourage physicians from using genuinely beneficial AI tools at all, purely out of self-protection. [1]

That finding matters because it shows the legal system has been actively working through both directions of this problem for years already. Not just “what happens if I trust the AI and it’s wrong.” Also, “what happens if I ignore the AI and a reasonable physician in my position would not have.”

A 2024 Milbank Quarterly analysis from the Milbank Memorial Fund reinforces the same structural point. A physician who relies in good faith on an AI/ML recommendation can still face liability if the actions taken fall below the standard of care. Liability does not disappear because a tool was involved. It relocates around the tool. [5]

And a 2026 focus-group study conducted through the CLASSICA project at the University of Illinois, led by law professor Sara Gerke, surveyed eighteen U.S. and European surgeons directly on this question. The surgeons largely agreed that ultimate responsibility for outcomes remains theirs, even when AI is involved in the decision. They do not yet consider AI part of the current standard of care. But, notably, many of them expect that it will be. [4]

Read that finding again. These are not outside commentators theorizing about physicians. These are practicing surgeons, on the record, expecting the standard they are held to will include AI before long.

V. Why This Should Matter First to Concierge and Membership Physicians

It would be easy to read all of this as a large-hospital, high-acuity, malpractice-attorney problem. It is not only that.

Concierge and membership medicine was built on a promise of higher engagement, more time, deeper relationship, and better outcomes than the volume-driven system most patients left behind. That promise is real, and CMT has documented it for nearly two decades.

But first principles cut both ways. A model built on the claim of being more attentive, more current, and more personally invested in each patient’s outcome is also a model that will be judged, fairly or not, against a higher expectation of staying current. A membership practice charging patients directly for a more engaged standard of care is in a weaker position than most to argue later that it simply had not gotten around to the tools its own patients were already using at home.

This is not a reason for alarm. It is a reason for posture. The physicians best positioned for whatever the legal landscape becomes over the next decade are not the ones scrambling to adopt a tool after a case makes headlines. They are the ones who can already explain, clearly and specifically, how they use AI, where they trust it, where they overrule it, and why.

VI. What This Does Not Mean

It does not mean physicians should adopt every AI tool that reaches the market. Multiple sources cited here make the opposite point directly. Physicians are still expected to exercise independent clinical judgment, not defer to AI outputs blindly, and document that judgment clearly. [9] Courts evaluating an AI-related claim have been described as likely to weigh three specific factors: whether the physician understood the AI tool’s limitations, whether the AI’s recommendation significantly departed from established guidelines, and whether the physician exercised independent clinical judgment in interpreting it. [2]

In other words, the emerging standard is not “did you use AI.” It is closer to “did you engage thoughtfully with the tools available to a reasonable physician in your position, and can you explain the judgment you exercised either way.”

That is a demanding standard. It is also, not coincidentally, exactly the kind of standard relationship-driven medicine was already built to meet.

VII. Learn, Build, Lead

Learn. Physicians do not need to become AI engineers. They need working literacy in what the tools available in their specialty can and cannot do, and where the documented failure points are.

Build. Practices should be documenting AI use, and non-use, as a matter of course. Not defensively. Deliberately. A brief, honest note on why a tool was or was not used in a given clinical decision is the same discipline good physicians already apply to every other judgment call they make.

Lead. The physicians who will be least exposed as this standard shifts are not the ones who adopted first. They are the ones who can articulate their reasoning clearly, to a patient, to a colleague, or eventually to a jury.

A Final Word

None of this is a prediction of what a court will decide next year, or in five years. Legal scholars themselves note there is still little case law directly on point, and that courts, regulators, and insurers are only beginning to work through what accountability looks like in this era. [9][4]

But the physicians who shape what comes next will not be the ones who waited for the first headline-making verdict to start paying attention. They will be the ones already asking the question this article is asking now, quietly, before anyone requires them to.

This article is published for educational and informational purposes only. It does not constitute legal, medical, or professional advice. Physicians should consult their own malpractice counsel and carrier regarding their specific practice. © 2007-2026 Concierge Medicine Today, LLC. All rights reserved.

Sources & Citations

• [1] Tobia, K., Nielsen, A., Stremitzer, A. “When Does Physician Use of AI Increase Liability?” Journal of Nuclear Medicine, January 2021. DOI: 10.2967/jnumed.120.256032. ncbi.nlm.nih.gov/pmc/articles/PMC8679587
• [2] “The New Standard of Care? AI and the Future of Medical Malpractice Law.” Suffolk University Journal of Health & Biomedical Law, January 2026. sites.suffolk.edu/jhbl
• [3] “Liability for Use of Artificial Intelligence in Medicine.” Research Handbook on Health, AI and the Law, NCBI Bookshelf, 2024. ncbi.nlm.nih.gov/books/NBK613216
• [4] “The New Malpractice Frontier: Who’s Liable When AI Gets It Wrong?” Medical Economics, citing Sara Gerke, JD, University of Illinois Urbana-Champaign, and the CLASSICA Project (Horizon Europe), published in Annals of Surgery Open. medicaleconomics.com
• [5] “Artificial Intelligence and Liability in Medicine: Balancing Safety and Innovation.” Milbank Quarterly, Milbank Memorial Fund, April 2024. milbank.org
• [6] “Implications of Current Litigation on the Design of AI Systems for Healthcare Delivery.” arXiv preprint, 2025, citing Rock Health physician survey data. arxiv.org/pdf/2507.15981
• [7] “Algorithmic Authority and the Clinical Standard of Care.” arXiv preprint, 2026, citing Chew et al. 2025 and Solow-Niederman 2025. arxiv.org/pdf/2606.00044
• [8] “AI In Medicine And Medical Malpractice: A New Landscape.” McCune Wright Arevalo LLP, July 2025. mccunewright.com
• [9] “AI in Medical Malpractice: Liability, Risk, & What Doctors Need to Know.” Indigo, June 2026. getindigo.com
• [10] “‘Physicians Who Use AI Will Replace Those Who Don’t,’ AMA Chief Says.” Becker’s Hospital Review, citing AMA President Jesse Ehrenfeld, MD, July 2023. beckershospitalreview.com
• [11] “AI Is Already Reshaping Care. Here’s What It Means for Doctors.” American Medical Association, citing Dr. Jesse Ehrenfeld at the ViVE health technology conference, April 2024. ama-assn.org
• [12] “AMA: AI Usage Among Doctors Doubles as Confidence in Technology Grows.” American Medical Association, March 2026. ama-assn.org
• [13] Whyte, John, MD, MPH. “AMA CEO: AI Won’t Replace Doctors, It Will Work Alongside Them.” STAT News, September 2026. statnews.com
• [14] “Using AI to Heal the System, Not Replace the Physician.” American Medical Association, citing Dr. Baligh Yehia, Jefferson Health, September 2025. ama-assn.org