AMA seeks laws to shield doctors from AI liability

Closing the liability gap

The American Medical Association (AMA) recently adopted a policy to advocate for new laws regarding artificial intelligence in healthcare. The group wants to ensure that physicians are not held legally responsible when they choose not to follow an AI tool's recommendation. This move comes as health systems increasingly integrate AI into areas like prescribing, diagnosis, and care management.

Currently, malpractice law generally focuses on whether a doctor met the standard of care. However, legal scholars have warned that doctors could face blame for both following a flawed AI suggestion and for overriding a correct one. The AMA wants to protect independent clinical judgment so doctors can exercise oversight without fear of litigation.

As Medical Daily reported, the AMA's goal is to settle the question of accountability before these tools become routine in hospital wards. AMA President Dr. Willie Underwood III stated, "Patients deserve to know who is responsible for their care." He emphasized that doctors need the authority and protection to make independent decisions when AI is part of the care team.

Shifting blame to vendors

A core part of the AMA's proposal is shifting the burden of liability from the doctor to the company that created the software. The policy suggests that when a defective clinical AI product harms a patient, the responsibility should rest with the vendor, manufacturer, or distributor. This distinction is intended to ensure that those who build the tools are accountable for their technical failures.

The AMA is also calling for legislation that requires AI companies to carry adequate liability insurance. They want these companies to be transparent about their product limitations and performance data. Furthermore, the association wants to prevent contracts from shifting the responsibility for product failures onto the doctors or the health systems that use them.

The policy also addresses organizations that mandate specific AI tools for their staff. If a hospital or payer requires a specific tool and that tool causes harm, the AMA argues that the mandating organization and the vendor should share the liability. This could be particularly important for doctors employed by large systems that roll out AI across all their clinics.

Regulations and licensure

Beyond liability, the AMA is seeking specific regulatory steps to build trust in medical AI. They are calling for model legislation that requires transparent, auditable evidence of safety and effectiveness for any AI used in diagnosis or prescribing. They want to ensure that the tools are being used as intended and that their outputs are consistently reliable.

The association is also taking a firm stance on the professional status of AI. Delegates stated that AI cannot be granted physician licensure, professional titles, or certification to practice medicine. They are calling for laws that would bar nonhuman entities from being presented with titles that are reserved specifically for human physicians.

While the AMA's policy is advocacy and not current law, it sets a clear public record for the nation's largest physician group. It does not currently change malpractice rules in any state. However, it provides a framework for how the group will lobby for changes in statehouses and in Congress as the technology matures.

Current AI adoption trends

The urgency of these legal protections is underscored by how quickly AI is entering the medical field. A survey of nearly 1,700 physicians found that 81% of them reported using AI in their practices. This figure is more than double the share reported in 2023, showing a rapid increase in adoption.

Currently, many doctors use AI for administrative and research tasks, such as summarizing research papers or drafting discharge instructions and care plans. However, tools that assist with actual diagnosis are also beginning to spread. Physicians in the survey ranked the establishment of clear liability frameworks as the most important regulatory step needed to build trust.

Until new laws are passed by state legislatures or Congress, courts will continue to apply existing malpractice and product liability laws on a case-by-case basis. This means that legal outcomes may vary significantly from state to state. The AMA's policy serves as a roadmap for how they want those legal standards to be standardized across the country.

What patients can do

Patients do not need to change how they seek care, but they can be proactive in asking questions about the technology used in their treatment. They can ask if AI tools are being used for visit notes, test results, scans, or treatment suggestions. They can also ask whether a human clinician reviews the output and who to contact if they have concerns about a specific decision.

For major decisions like surgery or cancer treatment, patients can request a second opinion to ensure a human clinician's judgment is being prioritized. If a patient receives an insurance denial, they can ask if an automated tool was involved and whether a licensed clinician reviewed that decision. They also have the right to file an appeal in these instances.

It is important for patients to keep records of their medical history and notes on conversations regarding major treatment decisions. Finally, the AMA notes that anyone experiencing emergency symptoms like chest pain or trouble breathing should call 911 rather than relying on an app or chatbot for guidance. Keeping a clear record of who made which decisions remains a key way to ensure accountability.