Utah just handed an AI chatbot the keys to part of its prescription system.
In a first-of-its-kind pilot, the state is allowing artificial intelligence to autonomously approve medication refills for residents—no doctor in the loop, at least not after the initial testing phase.
Public health advocates have a blunt response: this is “dangerous.”
How the Utah AI refill program works
The program runs through Utah’s “regulatory sandbox,” a framework that lets companies trial “innovative” products while some state rules are temporarily waived.
The Utah Department of Commerce has partnered with Doctronic, a telehealth startup built around an AI chatbot branded as an “AI doctor.”
Today, Doctronic offers a nationwide flow like this:
- Patients chat with the AI doctor for free.
- If they want to proceed, they can pay $39 to book a virtual visit with a physician licensed in their state.
- Crucially, you must go through the chatbot before you ever see a human.
According to a non–peer-reviewed preprint from Doctronic, based on 500 telehealth cases on its platform, the company says:
- The AI’s diagnosis matched a clinician’s diagnosis in 81% of cases.
- Its treatment plan was “consistent” with a doctor’s in 99% of cases.
Those are headline numbers—but they haven’t been vetted in a medical journal.
What changes for patients in Utah
For Utah residents, the AI is now allowed to go a step further: it can refill existing prescriptions on its own.
Here’s the flow, as described in the pilot:
- A patient signs in and verifies they live in Utah.
- The chatbot pulls their prescription history.
- It surfaces a list of medications eligible for renewal.
- The patient can request a refill, which the AI may approve—for a $4 service fee.
Politico reports that the system can handle refills for 190 common medications used to manage chronic conditions.
Some key drug categories are excluded:
- Pain medications
- ADHD medications
- Injectables of any kind
So this is not a blanket refill engine for every drug on the market. It’s tightly scoped to a specific set of chronic-care meds, at least on paper.
Human review—then the AI is on its own
Utah and Doctronic have built a ramp-up period into the program.
For each class of medication, the first 250 refills approved through the chatbot will be reviewed by human doctors. After that threshold, the AI makes the call alone.
Adam Oskowitz, Doctronic co-founder and a professor at the University of California, San Francisco, told Politico that the system is designed to be conservative. When the bot isn’t confident, it is supposed to escalate the case to a real doctor.
“Utah’s approach to regulatory mitigation strikes a vital balance between fostering innovation and ensuring consumer safety,” Margaret Woolley Busse, executive director of the Utah Department of Commerce, said in a statement.
The bet from Utah regulators is clear: use the sandbox to safely push the boundaries of AI in healthcare, while still putting some guardrails in place.
The FDA question
Whether federal regulators agree is an open question.
On one side of the legal line, prescription renewals are part of practicing medicine, which is largely handled at the state level.
On the other, the Food and Drug Administration has said it has authority over medical devices used to diagnose, treat, or prevent disease—including software.
That raises a looming issue: at what point does an AI prescriber become a regulated medical device, rather than just a tool inside a state-run experiment?
For now, it’s not clear if or when the FDA will step into the Utah pilot.
“A dangerous first step”
Not everyone is willing to wait and see.
In a statement, Robert Steinbrook, who leads the health research group at watchdog organization Public Citizen, sharply criticized both Doctronic’s model and the lack of federal oversight.
“AI should not be autonomously refilling prescriptions, nor identifying itself as an ‘AI doctor,’” he said.
“Although the thoughtful application of AI can help to improve aspects of medical care, the Utah pilot program is a dangerous first step toward more autonomous medical practice,” he added. “The FDA and other federal regulatory agencies cannot look the other way when AI applications undermine the essential human clinician role in prescribing and renewing medications.”
That’s the core fault line here: is this a smart use of automation to cut telehealth costs and reduce friction for chronic patients—or the start of a slow drift away from human clinicians making core prescribing decisions?
Utah has made its choice, at least for now. The rest of the US—and its regulators—will be watching what happens as an AI doctor starts signing off on refills, one $4 click at a time.
