Loading page…
The program covers acne care, pelvic-health therapy and prescription renewals. Agreements with two major health systems create a path for further pilots—not automatic permission to deploy them.
Listen to this story
Utah’s three pilots test distinct, limited clinical workflows: acne care, remote pelvic-health therapy and refills for existing non-controlled prescriptions. Each begins under a 12-month agreement, and six independent evaluators will assess safety and effectiveness; Utah retains the authority to approve participants and decide on renewals. The state’s agreements with Intermountain Health and University of Utah Health may enable future proposals, not automatic testing. Results are expected to inform legislation on privacy, transparency and licensed clinicians’ roles, while liability and the limits of AI decision-making remain unsettled.
August AI says its initial refill program covers about 148 medications; it will not start prescriptions, change doses or substitute drugs.
Nolla Health’s Utah-only app lets users proceed without waiting for its clinician team, with escalation routes for more serious acne cases.
Expect Fitness plans to use AI for intake, treatment plans and follow-up so licensed therapists can serve more pelvic-health patients remotely.
Utah is giving AI a larger role in patient care under temporary regulatory exceptions. On October 5, 2026, officials added a health care pillar to the state's Pro-Human AI Initiative, announcing three clinical AI pilots and independent evaluation networks. The pilots cover acne treatment, pelvic-health physical therapy and renewals of existing prescriptions.
Utah's regulatory sandbox lets companies seek limited, temporary relief from specific state rules while testing products under conditions negotiated with regulators. Before a pilot proceeds, the Office of Artificial Intelligence Policy requires plans for patient safety, consumer benefit and data reporting. Participating companies must also disclose when AI is running the process.
New master agreement frameworks with Intermountain Health and University of Utah Health address data privacy and AI governance. They allow departments within those systems to propose future pilots, but each proposal still needs state review and approval. The agreements do not give every department permission to begin clinical AI testing.
Commerce Director Margaret Woolley Busse said the aim is better access, lower costs and greater clinician capacity. She also pointed to geographic barriers and the need to stretch a scarce health care workforce.
University of Utah Health innovation chief Dr. Jim Hotaling offered an example of AI assisting a clinician. He said a tool found a buried family-history detail: three male relatives of a young patient had died before 40 from sudden cardiac death. Hotaling confirmed the history with the patient, who was then evaluated and treated for a potentially lethal heart rhythm problem.
The pilots begin with 12-month agreements. Companies must notify the state before starting operations, and Utah can consider renewals afterward. For August's program, the state also cited clearer liability provisions and improved integration with electronic health records.
Clinical tools are higher stakes and they need stronger evidence to earn the trust of health systems and patients.
Mark Sendak, CEO of VegaHealth
Six independent evaluators will assess evidence from the pilots for safety and effectiveness. Sendak said that review will examine expected benefits, possible ways a product could fail and evidence that could change an approval decision. Utah retains authority over which companies enter the sandbox; evaluators do not replace the state regulator.
Alongside Vega Health, initial evaluators include Clarion AI Partners, the Coalition for Health AI, Glacis Technologies, Empathic Health and Stanford's Clinical Excellence Research Center.
Prescription renewals have already drawn scrutiny from Utah's Medical Licensing Board. Its concerns about the earlier Doctronic pilot included side effects, drug interactions and whether continued treatment remains appropriate. Board member Bill Hamilton said its working group is not a blanket endorsement of AI pilots, emphasizing physician-led oversight, phased safety review and patient disclosures.
Busse said the initiative will examine legal liability for doctors involved in AI-assisted care. Officials have not yet defined which clinical decisions AI should never make. Busse said they still want a human involved somewhere, while acknowledging that the boundary remains unsettled.
Lawmakers said pilot results will inform future legislation on patient privacy, consumer transparency and the role of licensed professionals in AI-assisted care.
Story updates
Utah is adding outside scrutiny to health AI experiments without putting the state on the hook for evaluation bills. On October 5, 2026, its Office of Artificial Intelligence Policy announced agreements with six organizations to support oversight of sandbox pilots. The companies being reviewed will pay evaluators directly, under requirements for independence, safety reporting and state audits.
Companies in Utah’s AI Learning Laboratory already report on their own performance. The new arrangements add checks before approval and during operation, rather than leaving the office dependent on company accounts. Utah’s evaluator framework separates that work into clinical and technical evaluation, qualifying other evaluators, and recording whether specified safety checks occurred.
Clinical and technical evaluation: Stanford Clinical Excellence Research Center, mpathic and Vega Health can assess AI systems before and during pilots. This work can include clinical review, testing for unsafe behavior and auditing live activity.
Evaluator network: The Coalition for Health AI qualifies reviewers from its partner program. It does not evaluate pilots itself, and its qualified reviewers face the same independence rules as other evaluators.
Runtime attestation: Clarion AI Partners and Glacis Technologies issue signed digital receipts showing that specified safety checks ran. Glacis uses the openly published OVERT standard for those records.
A clinical evaluator can examine privacy protections and have medical experts review a proposal. Once a pilot starts, it can sample the AI’s actual activity and compare it with the company’s reports. That gives the office a way to check reported performance against what the system did.
No evaluator has an exclusive role. A company directed to obtain an evaluation can choose any evaluator approved by the office. The company pays that evaluator directly; Utah does not fund the review.
Utah pairs that payment structure with reporting duties to the state. Evaluators must certify that each evaluation was independent, report patient-safety concerns whether or not a company passes, and let the office audit their work. A certification that contradicts actual findings can be grounds to pause, roll back or terminate a pilot.
The agreements do not themselves grant regulatory exceptions or endorse an evaluator or pilot. Decisions about whether a pilot starts, continues or stops remain with the office. Each agreement runs for 12 months, with annual renewal specified for the clinical evaluators and the Coalition for Health AI.
The same October 5 announcement included pilots for Nolla Health’s acne care, Expect Fitness’s pelvic-health physical therapy and August AI’s prescription refills. Utah says the refill pilot covers routine, non-controlled chronic-disease medications and requires physician review before prescriptions are authorized. Umbrella agreements with University of Utah Health and Intermountain Health create review pathways for future, individually vetted pilots.
The oversight framework is still taking shape. The office says evaluation criteria are being finalized with each organization. Medical Daily reported that evaluator fees and assignments to individual pilots have not been disclosed. Utah says data gathered through the healthcare pilots will be shared with the Legislature and the public to inform permanent regulations.
utah.govLoading discussion...
Join the conversation
Explain whether controlled trials should help draw those boundaries or wait for them.
Be the first to share a perspective or an experience.
Reader comments
Newest comments first. Replies stay oldest first.