Utah pilot lets AI write acne prescriptions with no doctor's visit

What happened

Utah has allowed an AI program to write prescriptions for acne medicine without a doctor writing them, as first reported by Medical Xpress (Medical Xpress). The report, which relays the LA Times, says New York City-based Nolla Health began offering AI-generated prescriptions to Utah patients on Oct. 5.

The scope is narrow: eight topical acne medicines, not pills. Utah set aside its licensed-prescriber requirement for the pilot. Nolla CEO Luis Wenus told the LA Times that each prescription will still carry a clinician's name, even though no person writes it.

Medical Xpress calls Utah the first state to do this. That seems to mean writing new prescriptions. Utah's earlier AI pilot, from a company called Doctronic, renews existing ones.

What the reports say about oversight

Human review is staged, and it loosens as the patient count grows.

On cost, the pitch is about $5 a month for the app and about $50 a month for compounded medicines. We did not verify typical dermatologist visit prices. Those figures come from the company's pitch as relayed by the news item, not from independent pricing data.

The American Medical Association is skeptical. CEO John Whyte said AI tools can help but should not practice medicine on their own. On whether they can meet that bar, he said: "I don't think they can meet that burden."

What this does not show

The biggest gap: no one has reported whether this works or is safe. No clinical safety or effectiveness outcomes have surfaced for the pilot.

• We have not seen the agreement text, so reporting requirements and escalation rules are unknown.
• Eligibility is unclear, including age limits, pregnancy, and which of the eight drugs a patient might get.
• How the AI assesses acne has not been verified. Some summaries mention facial scans, but we could not confirm that.
• Liability for harm is unaddressed in what we read, as is what the clinician-name arrangement means in practice.
• The reporting does not say what oversight looks like past the first 500 patients.

The wider acne research does not fill the gap. A 2025 scoping review says AI may allow earlier acne diagnosis and treatment, but bias in training data remains a challenge for health equity. We read only the abstract, and we found no outcome data for AI-issued acne prescriptions.

What it means for you

First, the Gen Z angle comes with a limit. Acne is common in teens and young adults, but nothing we read says who qualifies for Nolla's pilot. The earlier Doctronic pilot was adults-only. We do not know whether Nolla has an age cutoff.

The drugs are topicals, not pills, which narrows the stakes. It does not tell you how well the AI performs, and nobody has published that yet.

If you are in Utah and curious, a few questions are worth answering before you sign up:

• Who is the named clinician, and can you reach a person if something goes wrong?
• Is a physician reviewing prescriptions before they go out, or only afterward?
• What does the service do if you report a bad reaction or no improvement?
• What happens to your photos and medical history?

If you are unsure what a prescription is or whether it suits you, a pharmacist or clinician is a reasonable second check. The low monthly price is a pitch, not proof of quality.

Context: how Utah got here

Utah's earlier pilot is Doctronic's renewal agent, announced in January 2026. According to a JAMA Health Forum analysis by Stanford's Michelle Mello, Utah agreed not to enforce its unprofessional-conduct laws against Doctronic if the company follows a contract with safety and privacy protections. It covers 192 chronic-condition drugs, for adults only. We have not confirmed that Nolla's pilot uses the same mechanism.

Doctronic's guardrails were similar in shape: physicians review the first 250 patients before the AI acts, then the next 1,000 retrospectively. Mello noted that the proposal described no pretesting of the prescribing agent. The company pointed to a non-peer-reviewed preprint written by its equity owners, in a different population, that measured physician agreement rather than AI versus physician care. She recommended independent evaluators and published metrics.

Not everyone in Utah was on board. The state's Medical Licensing Board protested that it was not consulted and warned of risk to residents. The state kept the program but dropped two drugs from the list.

Federal law is a separate question. A Medical Xpress summary of a NEJM Perspective quotes a legal point that state waivers do not automatically make unproven medical technology lawful federally. The proposed Healthy Technology Act of 2025 would amend federal law to let AI prescribe.

Other states are watching. Lawmakers in Iowa, Idaho and elsewhere have introduced bills to license AI medical services, many based on a Cicero Institute template. JAMA (The First AI Drug Prescriber) and JAMA Internal Medicine (When an Algorithm Renews a Prescription) have also published commentary. We saw only the titles.