AI in Patient Care: Managing Malpractice Risk
October 20, 2026
byby Valerie Grove
Editor’s Note: The following opinion article was contributed by Valerie Grove, a medical malpractice attorney with Joseph Greenwald & Laake. As states increasingly enact laws requiring disclosure of artificial intelligence use in patient care, questions surrounding medical liability remain largely unresolved. Grove examines the emerging legal risks associated with AI-assisted diagnosis and documentation, the responsibilities of physicians using these technologies and the potential implications for medical malpractice litigation. The views expressed are those of the author and do not necessarily reflect those of Medical Liability Monitor.
In primary care and emergency departments, the use of ambient scribes – artificial intelligence tools that record and transcribe the audio of medical visits – has become standard practice, and other AI diagnostic tools, such as radiology imaging overreads, are emerging for use in the medical field as well. As these tools become more common, they will introduce new risks for physicians when errors occur. At this point, AI has only been at the center of a few medical malpractice cases, so the legal framework around these cases is still evolving – potentially exposing physicians to liability.
Where Ambient Scribes Create Risk
While ambient scribes are helpful, they have limits. Most obviously, they can’t see or hear the physical exam, so if the physician doesn’t voice their observations, the note will have no record of it.Similarly, a patient’s nonverbal cues are not captured or considered by AI scribes. Limping or visible pain will not be noted in the transcript unless the provider adds it while reviewing the note.
There is also an understated risk embedded in the way these tools are designed to function. AI is a statistical prediction tool, so it doesn’t flag ambiguous audio for uncertainty. Instead, it resolves the ambiguity independently, often inventing data and findings to resolve its uncertainty or question without notifying the provider it did so, potentially resulting in a hallucination that then enters the medical record.
An additional layer of risk arises once physicians get used to AI. Studies suggest that when physicians become comfortable with these tools, they rely on them more, which leads to more errors. This overreliance without verifying is the leading malpractice risk with the use of scribes today.
Informed Consent Remains Untested
Several states have enacted laws requiring physicians to disclose the use of AI tools – and paperwork that simply says a hospital “may use AI” is unlikely to withstand a legal challenge as to the sufficiency of informed consent.Legally, informed consent is defined as what a reasonable healthcare provider should disclose to a patient regarding a particular risk. Since the use of an AI tool in patient care creates risk, it needs to be properly disclosed. Simply Including a statement in an intake form that AI may be used likely will not be sufficient informed consent since it doesn’t inform the patient about what the tool does, the risks to the patient, its limitations nor that ability for the patient to “opt out” of the use of AI for his or her clinical care. Providers should, at a minimum, communicate with the patient about the use of the tool and the potential risks involved while clarifying to patients that the technology does not replace their own exam, assessment and diagnosis.
While including this type of disclosure may satisfy the requirements of informed consent, it would, of course, be fact specific, and AI informed consent requirements have not been tested in court.
A Tool in the Eyes of the Law
With little case law in this area to date, the law treats AI like any other instrument a physician uses. Any shift in this approach is likely to be narrow and tied to specific facts – for example, proof that no reasonable physician in similar circumstances would have relied on a particular output such as an imaging result that they would have recognized as wrong on review.Who Gets Named in a Claim
When a patient becomes the victim of an AI error, the physician who delivered their care becomes the target of the medical malpractice claim, not the AI software company. Because the patient has no direct contractual relationship with the AI company, any claim against the software company would not likely survive a motion to dismiss. Unlike a product liability case where, for example, a manufacturer of a defective toaster would be held liable because the product was marketed and sold directly to consumers for their use, AI software is used by the physician rather than his or her patients.Best Practices for Physicians
The guidance for physicians is clear: Proceed with extreme caution and review every output, every time. An AI diagnostic result is not a replacement for a consultation with a specialist; and that initial AI output still requires an independent review. Physicians must continue to apply their training, experience and judgment and avoid overreliance on artificial intelligence that can introduce errors.Physicians should read their professional liability policies and ensure they understand their coverage. Carriers are regularly rewriting terms in response to AI and the changes it’s bringing. Some policies contain clauses that exclude coverage for incidents involving certain AI situations while others require a separate rider for AI coverage. Providers should confirm with their broker or agent that their policies cover AI hallucinations and negligence actions that sound in AI errors.
The law around AI and malpractice is still developing, but for now, the risk rests with the physician. Review every output, disclose AI use to patients and confirm coverage before a claim forces the question. Errors will happen, but preparation can prevent a manageable incident from becoming a costly one.
Valerie Grove is a principal at law firm Joseph Greenwald & Laake, where she focuses her practice on representing injured victims of medical malpractice using her 20 years of experience litigating medical negligence cases. She previously represented providers in malpractice litigation and related health law issues. She can be reached at vgrove@jgllaw.com.
Leave a Reply