RFK Jr. cites Sam Altman to claim that not using AI in the exam room is now 'malpractice': OpenAI has not confirmed the remark

By 13WHAM (Sinclair) · September 29, 2026.

Robert F. Kennedy Jr., US Secretary of Health and Human Services (HHS), said at a summit of the MAHA ("Make America Healthy Again") movement in Washington that he met this week with Sam Altman, CEO of OpenAI. As Kennedy told host Armstrong Williams in an interview, Altman confessed to him that two years ago he would not have advised anyone to ask ChatGPT for medical advice, but that "at this point it would be malpractice for a doctor not to consult" those tools. Kennedy used the anecdote to argue that "medicine is going to change completely" and that Americans must "become the CEOs of their own health."

In the same appearance, Kennedy credited his tenure with a leap in the sharing of medical records: when he took office, he says, barely 10 million medical records were being shared openly across the entire sector; by June, some 15 months later, the figure had reportedly reached one billion, thanks to the removal of "barriers." Kennedy himself framed the MAHA summit as the meeting point between his agency, hospitals, pharmaceutical companies and tech firms around that same transformation.

These two figures deserve the same caution with which any politician's statement about his own record should be read. Neither the quote attributed to Altman nor the jump from 10 million to one billion records has independent public confirmation in the source: there is no record of OpenAI corroborating those words, and no audit is cited for the interoperability figure. And "malpractice" is not just any rhetorical phrase: it is a legal term regulated in the US by each state, not by HHS, so the quote —if accurate— describes a personal opinion of Altman's, not a change in the standard of care.

That is where it is worth distinguishing between "consulting" and "replacing clinical judgment." Today's models still produce documented factual errors in medical contexts, and neither the FDA nor any other regulator has approved the general use of a chatbot as an autonomous diagnostic tool. A model being useful as a quick second opinion —summarizing records, suggesting differential diagnoses, translating clinical jargon for the patient— is not the same as the regulatory validation that the word "malpractice" suggests.

And here the messenger matters as much as the message. Kennedy is the same secretary who has pushed changes to the childhood vaccination schedule that the scientific community has heavily criticized, and who for years defended discredited theories about vaccines and autism. That this is the voice claiming authority to certify which AI advice deserves clinical trust, leaning on an unverified quote from Altman, is exactly the pattern to watch: borrowed tech enthusiasm used to shield positions that already existed before AI entered the conversation.

That said, the substance of the matter —patients with full access to their own records, models that help interpret tests and symptoms, more personalized medicine— points in a direction we share over the long term: the more health data flows to the patient and to well-audited diagnostic support tools, the closer we are to healthcare that is predictive and not merely reactive. An abundance of health information, managed with the right safeguards, is part of the same path that can shorten diagnoses and extend healthy years of life. But that path is built with verifiable clinical standards, not with hallway anecdotes repeated on a television set.