Utah Green-Lights an AI That Prescribes Acne Medication for $5 a Month
Utah startup Nolla Health is running a one-year pilot in which a $5/month AI scans faces, scores acne severity, and prescribes from eight approved treatments — with physician oversight shrinking over time.
Updated
Why it matters
- Nolla Health charges $5 per month for an AI service that scans a patient's face and can prescribe from a list of eight approved acne treatments.
- The Utah pilot runs for one year: the first 100 prescriptions get physician oversight, then the AI fully treats 400 patients with weekly physician reviews, then monthly audits of at least 10% of prescriptions.
- Utah Medical Association CEO Michelle McOmber said the association "never talked with them" and was not required to approve the pilot.
- Utah's Office of AI Policy, part of the state Department of Commerce, had the Utah Medical Licensing Board subcommittee review and comment on the plan.
- McOmber warned: "AI, if used inappropriately, can be very dangerous in healthcare."
A Utah startup has launched what it calls the first AI doctor that can prescribe medication — a $5-per-month subscription service that scans a patient's face, scores their acne severity, and issues prescriptions from a list of eight approved treatments.
Nolla Health, a healthcare startup focused on skincare and acne treatment, runs the service through an app. According to Bloomberg, patients scan their face and fill out forms covering their medical history. The AI software analyzes the scan, generates an acne severity score, and — if the scan and medical information warrant it — writes a prescription drawn from the eight approved treatments.
The program matters beyond dermatology. It is a live test of whether an AI system, not a human clinician, can hold prescribing authority in the United States — even in a narrow, low-stakes domain — and of what regulatory guardrails, if any, states will build around autonomous medical AI. Utah's Office of AI Policy, a body created to guide responsible AI policy in the state, sits at the center of that test.
What does the pilot actually involve?
The pilot is slated to run for one year, Bloomberg reports. It unfolds in three escalating phases:
- Phase one: The first 100 prescriptions will be overseen by a board-certified physician before they go to pharmacies.
- Phase two: If that works, the AI will fully treat 400 patients, with a physician reviewing the system's decisions weekly.
- Phase three: A physician will audit at least 10% of written prescriptions once per month.
The trajectory is the story. Human oversight shrinks from reviewing every single prescription, to weekly check-ins on 400 patients, to a monthly audit of a 10% sample. By the final phase, the overwhelming majority of prescriptions will leave the system without any individual human review.
A Nolla Health representative did not immediately respond to a request for comment.
Did Utah's medical establishment sign off?
Not exactly — and that is a significant detail.
Michelle McOmber, CEO of the Utah Medical Association, told CNET in an email that Nolla Health did not need the Utah Medical Association's approval to run the pilot program.
"We never talked with them," McOmber said. "The Office of AI Policy did have a short conversation with our current president, who is a dermatologist. OAIP also had the subcommittee of the Utah Medical Licensing Board look at [the plan] and comment."
The Office of AI Policy operates within the Utah Department of Commerce. Its stated role is to guide responsible AI policies in the state. According to McOmber's account, the office consulted the licensing board's subcommittee and the association's dermatologist president — but the state's main physician organization had no direct role in evaluating the program before launch.
McOmber draws a hard boundary around what is acceptable. She describes the acne application as "very limited and very specific" — fine, as far as it goes. But the Utah Medical Association "would not be in favor of it being used for anything else at this time as it was," she said.
Is the medical profession ready for autonomous prescribing?
The research picture is mixed. Studies report that doctors, surgeons, and mental health professionals rank among the professions least likely to be replaced by AI. Yet many medical professionals agree that AI's expansion into medicine is inevitable — whether as a tool assisting human physicians or as a handler of simpler tasks such as diagnosing acne and other skin conditions.
The disagreement is about pace, not direction.
"We think people are trying to go too fast in using and relying on AI in healthcare," McOmber said. "We do believe that AI, as a tool, can help physicians in different ways and that it will be used and even should be used."
McOmber's position leaves room for AI in the clinical workflow — just not in the decision seat. She says human physicians should always be involved in patient care, whether writing prescriptions or examining patients to diagnose conditions. In its current form, AI is best suited to simpler tasks such as gathering useful medical information and medical histories from patients, she says.
Her warning is blunt: "Care decisions ultimately should rest with the physician in consultation with the patient. AI, if used inappropriately, can be very dangerous in healthcare."
Why acne is the beachhead
The choice of dermatology is strategic, and the economics explain it. Acne is common, visually diagnosable, and treatable with a known set of medications — eight, in Nolla Health's case. At $5 per month, the subscription price undercuts virtually any conventional pathway to a prescription, including a primary care visit or a dermatology appointment.
A face scan plus a medical history form is a workflow that maps naturally onto computer vision and form-processing — the mature end of current AI capability. The severity score gives the system a quantifiable basis for its decision, which is also exactly the kind of artifact a physician can audit at scale.
The pilot's structure suggests both Nolla Health and Utah's regulators understand the sensitivity. Phase one keeps a board-certified physician in the loop for every prescription. The later phases bet that the AI's judgment holds up well enough that sampling — weekly reviews, then monthly 10% audits — is sufficient.
Whether that bet pays off will define the template. If Nolla Health completes its one-year pilot without incident, other states and other startups will have a working blueprint for autonomous AI prescribing in narrow domains. If it fails — a missed contraindication, a misread severity score, a patient harmed by an unaudited prescription — McOmber's warning about inappropriately used AI in healthcare will become the reference point, and the physician-in-the-loop requirement will harden into policy.
Original: nollahealth.com
More from Marcus Bennett
Show full bio
Senior reporter covering consumer brands and retail at AI In Context.
148 articles
Related articles
- Federal Documents Expose Failures of Medicare's AI Prior Authorization Pilot
- Google DeepMind Unveils AI Co-Clinician Research Initiative
- Blue Cross Estimates AI Hospital Coding Added $942 Million in Costs
- OpenAI Launches 'OpenAI for Healthcare' With HIPAA Support
- Google DeepMind Launches National AI Partnership With Singapore