Utah just became the first U.S. state to let an AI system examine a patient and write a prescription without a human doctor checking its work first.
The change, reported by TechSpot and discussed on Hacker News, allows an AI tool to carry out the kind of exam and prescribing decisions that have always required a licensed clinician's signoff. Until now, AI has mostly played a backstage role in medicine: drafting notes, flagging scans, or suggesting a diagnosis for a doctor to approve or reject. Utah's move removes that human checkpoint entirely for at least some cases. The report does not detail which conditions or drugs are covered, or how the state plans to handle mistakes.
That gap matters more than the headline. Every other serious deployment of AI in healthcare, from radiology to triage chatbots, has kept a licensed professional as the final word precisely because misdiagnosis and bad prescriptions carry real harm and real liability. Utah is testing whether that safeguard is optional, and the rest of the country will be watching the malpractice cases, not the press coverage, to find out.
Proponents will call this access for patients who cannot get a doctor's appointment. Skeptics will call it a liability experiment on the public. Both can be true. Either way, no human oversight is a line regulators elsewhere have avoided crossing, and Utah just stepped over it first.