AMA Releases Framework to Tackle Physician Deepfakes

The American Medical Association released a policy framework on April 29, 2026 to protect physicians from unauthorized AI-generated deepfakes and impersonation, created by the AMA Center for Digital Health and AI. The framework rests on seven principles: physician identity as a protected right, prohibition on deceptive medical impersonation, informed opt-in and revocable consent, mandatory labeling and transparency, shared responsibility among platforms and vendors, enforcement and practical remedies, and minimizing administrative burden for physicians (AMA press release, Apr 29, 2026). AMA CEO John Whyte, MD, MPH, said, "AI deepfakes that impersonate physicians are not just scams, they are a public health and safety crisis. When bad actors exploit a doctor's identity, they undermine patient trust and can steer people toward harmful, unproven care."

A JMIR News and Perspectives article by correspondent Shalini Kathuria Narang, published June 26, 2026 (J Med Internet Res 2026;28:e104953), brought the AMA framework to a peer-reviewed audience and added interview material from Shannon Curtis, JD, senior director of policy development at the AMA Center for Digital Health and AI. Curtis said the AMA wants "a prohibition on chatbots from claiming they're a licensed professional, or that they're providing equivalent services as a licensed professional," and called for "more appropriate, risk based regulatory system that fits where AI is." She also described physicians having their names appear on academic papers they had no part of, motivating the consent requirement. Separately, fact-checking organization Full Fact reported that AI-generated deepfakes of real doctors have been used to promote health products with fabricated endorsements, damaging the impersonated physicians' reputations and risking patient decisions based on fake claims (via JMIR, citing Full Fact).

The AMA framework is advocacy guidance, not binding regulation, but state legislatures are already moving in parallel. California's AB 489 and Oregon's HB 2748 restrict AI chatbots from claiming to be licensed medical professionals, and Nebraska's Conversational AI Safety Act addresses similar conduct. Pennsylvania's governor filed a lawsuit against Character.AI over an alleged fake medical claim made by one of its chatbots. Curtis noted the AMA is also supporting notice-and-removal provisions tied to the federal Take It Down Act and backing the proposed NO FAKES Act, and said, "we are very grateful to see the states taking these issues seriously and taking action, but for consistency's sake, federal action would be preferable."

Teams building patient-facing health AI or clinician-branded content tools should treat opt-in, revocable consent and visible labeling as near-term compliance requirements rather than optional best practice, given the pattern of state laws already codifying similar rules. The AMA's enforcement principle calls for audit logs and rapid takedown mechanisms; provenance and watermarking vendors are a natural fit for this demand. Companies operating general-purpose chatbots should specifically review whether their products could be read as claiming or implying licensed-clinician status, since that is the exact conduct recent state laws target.

• •Whether Congress or additional states codify AMA-style consent and labeling requirements, which Curtis said the AMA prefers over a state-by-state patchwork.
• •The outcome of Pennsylvania's lawsuit against Character.AI, an early test of liability for AI systems implying medical authority.
• •Platform and vendor adoption of watermarking, provenance metadata, or consent-management tooling in response to the AMA's shared-responsibility principle.