FAQs About AI in Radiology: Legal Risks, Liability, and Malpractice

Artificial intelligence (AI) utilization is growing within the health care space, from electronic health records to radiology, but what happens when AI makes a mistake or a provider disagrees with the tool’s decision?

Here are some common FAQs on the implications of using AI in radiology:

Who is legally responsible when AI is involved?

Current US malpractice law holds physicians accountable for malpractice regardless of whether AI was used or not. There is currently no federal statute that assigns shared responsibility to an AI system for all cases in which it is directly influencing patient care. However, multiple states are beginning to take varied approaches to regulating AI in health care: Oregon has prohibited AI from using protected clinical titles like “nurse”; Illinois has restricted the use of AI in therapy without licensed professionals; and states such as Texas are increasingly scrutinizing the role of AI in insurance decision-making.^1-3

Despite evidence of diagnostic delays and flawed outputs, physicians are often held responsible in AI-related malpractice cases, although liability can also extend to hospitals and is not uniform across cases.⁴

“Given what’s happened, it seems to me that software companies that develop AI tools will have a stronger hand in the process,” Brian Sheppard, JD, LLM, a professor of law at Seton Hall Law who focuses on medical malpractice litigation, said in an interview with The American Journal of Managed Care^® (AJMC^®). “I do suspect that they'll be in a better position to immunize themselves from liability.”

What happens if a radiologist disagrees with AI and the AI is correct, and vice versa?

Prior research shows that a jury was more likely to find a radiologist legally liable when they disagreed with the AI after their interpretation of evidence of a brain bleed or cancer, but the AI still flagged the pathology as abnormal. The same study found that jurors were also less likely to side with the patient in the case of an AI false discovery rate and false omission rate.⁵

Juror perception of AI plays a significant role in determining liability, and although they may not need to understand the intricacies of the tool, knowing the rate of error is a determining factor in the tool’s efficacy to meet or fall below the standard of care, Sheppard said.

“Telling juries simple things like false positive rates about AI tools can actually have a big influence on their perceptions of liability,” he said. “I do think having some basic command of strengths and weaknesses of a particular AI program will absolutely be relevant in the courtroom.”

How might jurors perceive AI in these cases?

Another recent study published in Nature Health found that jurors were more or less likely to side with the patient if the radiologist interpreted the CT scan twice: once without AI and then again with AI feedback.⁶ In this scenario, the jury found that the radiologist acted in accordance with the standard of care, using the degree of knowledge, skill, and judgment expected of a reasonably prudent radiologist under similar circumstances. Furthermore, jurors were also more likely to side with a radiologist if they were aware of the AI error rate.

“People are vulnerable to believing that AI is more accurate than it really is,” Michael Bernstein, MD, an assistant professor in the Department of Diagnostic Imaging at the Warren Alpert Medical School, said in an interview with AJMC. “If true, then patients might have unrealistically high expectations about what AI can accomplish, which could help explain why radiologists are viewed as more liable when they contradict AI in making a false negative [FN] interpretation relative to making an FN interpretation absent AI.”

Does workflow, or how AI is used, affect malpractice risk?

Research has shown that AI can increase radiologists’ workload and streamline their workflow, but there is still more to understand about how improving efficiency with AI influences the perception of liability.⁷

“Understand that liability is only one concern when you're engaging in the practice of medicine,” Sheppard said. “The number of reads, the order in which AI reads vs a human, the degree to which the human is aware of the results of AI or not—all of these things impact perceptions of liability.”

What legal standard is applied in AI-related malpractice cases?

Researchers and policymakers alike are calling for more reform when it comes to AI-related malpractice cases. For example, in aviation, when automation fails, the fault is distributed between the pilots, systems, and manufacturers. Health care systems have the potential to benefit from litigation reform similar to that of aviation, according to the Johns Hopkins Carey Business School.⁴

“Software companies do not typically have much exposure in these instances, leaving individual practitioners in hospitals holding the bag,” Sheppard said.

AI is increasingly used in radiology, but legal responsibility still largely rests with physicians. How radiologists interact with AI, agreeing or disagreeing with its findings, may shape future malpractice outcomes.

References

1. Artificial Intelligence 2025 legislation. National Conference of State Legislatures. July 10, 2025. Accessed March 19, 2026. https://www.ncsl.org/technology-and-communication/artificial-intelligence-2025-legislation
2. Gov Pritzker signs legislation prohibiting AI therapy in Illinois. Illinois Department of Financial and Professional Regulation. August 4, 2025. Accessed March 19, 2026. https://idfpr.illinois.gov/news/2025/gov-pritzker-signs-state-leg-prohibiting-ai-therapy-in-il
3. Creative E. From Colorado to Texas: How states are rewriting ai laws. Miller Nash LLP. December 4, 2025. Accessed March 19, 2026. https://www.millernash.com/industry-news/from-colorado-to-texas-how-states-are-rewriting-ai-laws

4. Fault Lines in health care AI—Part Two: Who’s responsible when ai gets it wrong? Johns Hopkins University Carey Business School. June 26, 2025. Accessed March 18, 2026. https://carey.jhu.edu/articles/fault-lines-health-care-ai-part-two-whos-responsible-when-ai-gets-it-wrong

5. Bernstein MH, Sheppard S, Bruno MA, Lay PS, Baird GL. Randomized study of the impact of AI on perceived legal liability for radiologists. NEJM AI. 2025;2(6). doi:10.1056/AIoa2400785

6. McCrear S, Bernstein M, Sheppard B. AI in radiology: how double-reading may affect malpractice risk and jury perception. AJMC. March 17, 2026. Accessed March 18, 2026. https://www.ajmc.com/view/ai-in-radiology-how-double-reading-may-affect-malpractice-risk-and-jury-perception

7. McCrear S. AI-assisted mammogram readings reduce radiologist workload and maintain performance. AJMC. August 20, 2025. Accessed March 18, 2026. https://www.ajmc.com/view/ai-assisted-mammogram-readings-reduce-radiologist-workload-maintain-performance