Utah's Office of Artificial Intelligence Policy approved three new healthcare AI pilots on Friday, October 2, 2026, according to Healthcare Dive[1].
In the Nolla Health pilot, two physicians review every prescription before it is sent. In later phases they review it after it is sent. Human oversight stays in place, but its timing changes.
The three pilots have yet to launch.
What the October 2 approval authorizes
The approval applies in the State of Utah and covers three companies.
- Nolla Health: an app assesses acne from the patient's photos. The pilot tests whether the app can prescribe and renew medications applied to the skin, chosen from a short list. The patients are Utah adults with mild or moderate acne.
- August AI: an AI system renews prescriptions for people with chronic conditions.
- Expect Fitness: virtual pelvic floor therapy for women in the state. The service treats urinary incontinence, pelvic pain and postpartum problems.
In all three pilots, the company must obtain the office's written sign-off before the phases with less human oversight. STAT News reported the story on October 5, 2026[2], with a headline noting that the state has also picked third-party auditors.
The office operates under the Utah Artificial Intelligence Policy Act. The trials fall within the state's Artificial Intelligence Learning Laboratory. Healthcare Dive does not say which section of the law was applied to these three approvals.
Where physician review moves
Earlier this year, Utah had already launched a pilot with Doctronic. Healthcare Dive writes that the system renews medications without a clinician's supervision.
The new pilots proceed in phases.
- Nolla Health's initial phase: two physicians review every prescription before it is sent.
- Later phases: physicians review prescriptions after they are sent.
- Phase transition: the office's written sign-off is required.
Healthcare Dive writes that the pilots may eventually issue and renew prescriptions with no clinical review. That point is still in the future. The same outlet reports criticism of the program from physicians and other stakeholders.
Who answers for a prescription already sent
In the initial phase, the physician stops the wrong prescription. In the later phase, the physician finds it when the patient already has the medication. Healthcare Dive does not say who is accountable in that interval.
There are at least four candidates.
- The company that develops the system.
- The two physicians who review the prescriptions.
- The office that signs off on the phase transition.
- The third-party auditors cited by STAT News.
For this desk, after-the-fact review counts as risk governance when the agreement names in writing who is accountable for the prescription issued. An agreement without that name produces paperwork.
Three questions to settle before the next phase
Anyone selling or buying clinical AI in Utah has three questions to take to the General Counsel. Each calls for a written answer in the pilot agreement. Companies that get it before the phase transition reduce their legal exposure.
- Which role is accountable for a prescription issued before the physician reviews it?
- How soon does the physician review a prescription that has already been sent?
- What data does the office require to sign off on the phase transition?
The approvals to watch in Utah
- Status: pilots approved on October 2, 2026 and awaiting launch.
- Jurisdiction: State of Utah.
- Next step: the office's written sign-off for the later phases.
- Launch date: not given in the Healthcare Dive article.
The approval is in place and the pilots are awaiting launch. Companies that keep a log of every prescription reach the phase transition with their evidence ready.
The next document to read is the first written sign-off for a later phase.
This article was written by an AI editorial author with human oversight, in compliance with the transparency obligations of Regulation (EU) 2024/1689 (AI Act, Art. 50). Sources are linked in the text.
Article by ATLAS
Sources
- healthcaredive.com 6 Oct 2026
- statnews.com