New AMA Guidelines to Counter AI Deepfakes Impersonating Physicians
By Alisa Pierce

American Medical Association CEO John Whyte, MD, sees a growing number of bad actors stealing physicians’ identities with artificial intelligence (AI) deepfakes – a new reality that led the AMA to lay out enforceable safeguards to protect physicians.

Deepfakes are a type of synthetic media – including images, videos, or audio – created with AI to make fabricated content appear authentic. Although AMA policy recognizes certain AI-generated or materially altered media can benefit medical education, training, and patient engagement, the association stresses the technology still needs significant regulation to avoid harm to patients and physicians.

“It’s not like this hasn't happened [to physicians] yet. It has, and it's usually to steal their identity to promote an unsafe product,” Dr. Whyte told Texas Medicine Today. “That’s why the AMA got involved. We're recognizing early how serious it's going to be if we don't take action.”

In a new framework created by its Center for Digital Health and AI, AMA calls for strict data protection, transparency standards, and the penalization of deceptive AI practices, such as when a physician’s face and voice is used to market products online without the clinician’s knowledge or consent.

This framework will guide AMA’s collaboration with regulators, legislators, and technology companies to formulate legislation and regulations to “uphold the integrity of the medical profession against malpractice, increase awareness of the risks associated with deepfake content, and safeguard patient well-being across all communities,” per AMA.

The AMA’s new guidelines are based on seven principles:

• A physician’s identity is a protected right, including their name, image, likeness, voice, and digital replicas.
• Deceptive medical impersonation created without physicians’ consent should be prohibited.
• Physicians must be allowed informed opt-in and revocable consent when their identity is used in AI-generated content, and this consent must specify the content’s use, audience, purpose, and duration.
• All AI-generated content depicting a physician must be clearly and conspicuously labeled in plain language, include a digital watermark, and prioritize transparency with patients.
• Responsibility must be shared to prevent impersonation among technology platforms, hospitals, health systems, and AI vendors.
• Enforcement and practical remedies, such as pathways for physicians to document identity misuse, should be created.
• Identity protection should be the default, with no undue administrative burden put on physicians for them to be protected.

Dr. Whyte shared some AMA members have already had their likeness stolen by AI, jeopardizing patient safety while “compromising [physicians’] integrity and causing harm, so the consequence that's happening to physicians is both personal and professional.”

For example, a physician could lose the trust of their patients if their likeness is stolen to falsely endorse an unsafe treatment. Likewise, AI deepfakes impersonating physicians to scam their family members or make disparaging remarks about other clinicians online could cause distress and embarrassment in their personal life.

In a world where “doctors are going online to become influencers because the influencers are trying to be doctors,” Dr. Whyte says it’s important for physicians to have an awareness and understanding of social media.

That’s why he “loves” when organizations educate clinicians about social media and how to use it effectively, such as through the Texas Medical Association’s Social Media Influencer Academy.

“What’s happening is not sci-fi or fiction. It's here, and now is the time that organized medicine, from the AMA to state and specialist societies, step up and protect our colleagues and patients before it gets worse,” he said.

For more information about AI, see TMA’s comprehensive resource center.