SALT LAKE CITY — Utah is expanding its artificial intelligence regulatory sandbox into the healthcare sector, announcing three new clinical pilots and partnerships with the state’s largest health systems. This move also includes establishing a network of outside evaluators designed to test whether the technology is safe and beneficial for patients.
This initiative marks a new healthcare pillar within Utah’s Pro-Human AI Initiative. It involves the Utah Department of Health and Human Services, Intermountain Health, University of Utah Health, medical licensing officials, and outside experts in the state’s AI policy process. The state’s regulatory sandbox allows companies to seek limited, temporary relief from specific state rules while testing AI products under conditions negotiated with regulators. The Office of Artificial Intelligence Policy requires companies to provide a plan for patient safety, consumer benefit, and data reporting before a pilot can move forward.
Commissioner Margaret Woolley Busse explained the goal is to improve healthcare access and affordability while preserving human responsibility for important medical decisions. Busse stated that if the state "can get this right," it could lead to better access, lower costs, greater clinician capacity, a reduction in geographic barriers, and more efficient use of a scarce healthcare workforce.
Key to this expansion are master agreement frameworks with Intermountain Health and University of Utah Health. These agreements cover critical areas such as data privacy and AI governance, allowing individual departments or divisions within each system to propose future pilots. These proposed pilots would still require state review and approval before proceeding.
Dr. Jim Hotaling, University of Utah Health Chief Innovation Officer and a practicing surgeon, provided an example of AI's potential utility. He described how an AI tool identified a critical detail—that three male relatives had died before age 40 from sudden cardiac death—buried in a young patient’s extensive medical record. After Hotaling spoke with the patient and confirmed the family history, the patient was evaluated and treated for a potentially lethal heart rhythm problem. Hotaling emphasized that the goal is "not to take the human out of the loop" but to "make the human a better human and better able to take care of patients."
The three newly announced pilots will take different approaches to clinical AI, beginning with 12-month agreements. Companies must notify the state before starting operations, and the state can consider renewals after the initial period. The state also announced a clinical AI advisory network and a third-party evaluation network. Mark Sendak, CEO of Vega Health, said evaluators will independently assess whether a product meets the state’s evidentiary standards, including its expected benefit, potential failure modes, and the evidence that could change an approval or rejection decision. The state, however, will retain authority over whether a company enters the sandbox. Initial evaluators include Clarion AI Partners, the Coalition for Health AI, Glacis Technologies, Empathic Health, and Stanford’s Clinical Excellence Research Center.
This expansion follows Utah’s earlier Doctronic prescription-renewal pilot, which uses AI to process certain renewals of medications already prescribed by a licensed provider. The state has indicated this pilot remains limited and subject to oversight. The Utah Medical Licensing Board had previously raised concerns about the prescription-renewal experiment, noting that refill decisions can involve side effects, interactions, contraindications, and the need to determine whether treatment remains appropriate. The board now has a working group supporting the AI policy office’s review of healthcare applications. Dr. Bill Hamilton, a physician and medical board member, clarified that this working group is not a blanket approval of AI pilots. He stated the focus is on physician-led oversight, phased safety review, patient disclosures, and gathering data to guide future licensing and healthcare policy.
State leaders acknowledged that they have not yet drawn a firm line defining which clinical decisions AI should never make. Commissioner Busse noted that the answer is not yet known due to the rapid changes in AI capabilities, but stressed that "we always want to have a human involved somewhere." Lawmakers anticipate that the results from these initiatives will help shape future legislation, particularly concerning liability, patient privacy, consumer transparency, and the role of licensed professionals when AI is involved in patient care.