Utah Lets AI Examine Patients and Prescribe Without a Doctor

Listen to this article

Narrated by Charlotte · The Noble House

Utah regulators have approved a pilot that lets Nolla Health's AI app issue initial acne prescriptions, with the company citing a 61-day average wait for a dermatology appointment in the state [5]. Patients complete an intake questionnaire and a five-angle face scan, and in the pilot's later stages the AI may prescribe without prior human review [1][8].

Utah has expanded its AI regulatory sandbox to approve new pilots of clinical AI tools for acne treatment, chronic medication renewals, and women’s health therapies [2]. By partnering with major systems like Intermountain Health and University of Utah Health, the state aims to establish a scalable model for integrating AI into routine care while attempting to mitigate the risks that have drawn criticism from doctors in the state [2].

The core of this expansion lies in the state’s willingness to grant temporary regulatory exceptions. These exemptions allow technology providers to test products under state supervision, effectively waiving certain licensing requirements for medicine and telehealth [4]. The OAIP has structured these pilots to include strict oversight mechanisms, including mandatory third-party audits and stage-gated progressions that require written regulator approval before companies can reduce human involvement [2]. This approach reflects a broader strategy to attract healthcare technology innovation while attempting to define the boundaries of safe AI deployment in a high-stakes environment [4].

Expansion of the Regulatory Sandbox and Provider Partnerships

Utah’s AI sandbox, established by state law in 2024, was designed to create a supervised environment where private companies could test AI tools before scaling them publicly [2]. The recent announcements mark a substantial increase in the program’s scope and complexity. The OAIP has signed master agreements with two of the region’s largest healthcare providers: Intermountain Health and University of Utah Health [3]. These agreements do not constitute pilots themselves but set the governance terms under which future pilots can be approved within these institutions [2]. This partnership allows the state to leverage the infrastructure and clinical expertise of major health systems to validate AI tools in real-world settings.

The involvement of these large providers signals a shift from isolated startup experiments to integrated healthcare workflows. Intermountain Health CEO Rob Allen described the sandbox as a structured approach to assure tool effectiveness before deployment, emphasizing patient safety [4]. University of Utah Health officials have expressed optimism about the potential for AI to address access gaps, particularly as Intermountain Health expects up to 40% of today's providers to retire by 2035 [3]. Dan Liljenquist, chief strategy officer at Intermountain Health, noted that current systems for training new physicians are not emerging quickly enough to meet future demand, positioning AI as a necessary bridge in the healthcare model [3].

The expansion also includes specific pilots with individual technology companies. Five entities received approval to use AI in settings that typically require state licensure [4]. These include Nolla Health for dermatology, August AI for prescription renewals, and Expect Fitness for pelvic floor therapy [2]. The state has also established a network of six independent third-party evaluators to oversee these pilots. Vega Health was selected as one of these evaluators, tasked with validating technical claims and auditing the accuracy of AI decisions before deployment [6]. This layer of independent verification aims to address concerns about reliance on vendor performance claims and to improve public trust in the deployed technologies [9].

Compass Predictive Analytics

Compass prediction

Forecast

Yes · Favor

Will "STAT+: Utah plows ahead with more health AI pilots for prescriptions, women’s health" produce verified policy implementation within 72h? Horizon 72h; target window 2026-10-07T06:05:49.487000+00:00 to 2026-10-10T06:05:49.487000+00:00.

Signal gauge

70%

Evidence Reliability

9 Of 9 Validated Assertions Have Complete Exact Span And Ownership Lineage. · Positive

tracked

Quantifies the conservative evidence floor after exact-span and independent-owner checks.

9 evidence references

Illustration: Utah regulators have signed master agreements with Intermountain Health and University of Utah Health to expand the AI regulatory sandbox, allowing major providers to test autonomous clinical tools under state supervision.

Clinical Pilots: From Acne to Chronic Care

The approved pilots cover a range of clinical tasks, each representing a different level of risk and complexity. The most notable is the approval for Nolla Health to issue initial acne prescriptions via AI without prior human review in later stages of the pilot [1]. Nolla Derm allows Utah residents aged 18 and older to complete an intake questionnaire and face scan to receive a treatment plan and prescription [8]. The service costs $4.99 per month during the pilot period, addressing access gaps where patients often face long wait times for dermatology appointments [5].

The Nolla Health pilot is structured in three stages of decreasing physician oversight. Initially, two licensed physicians review every AI-generated prescription before it reaches the patient [8]. In subsequent stages, the AI may prescribe autonomously, with physicians reviewing decisions retrospectively or sampling a portion of cases [8]. The company reports that clinicians agreed with the AI’s recommendations in more than 96% of real-world cases, with disagreements limited to minor adjustments in medication strength [5]. However, the transition from supervised to unsupervised prescribing represents a significant change in risk profile, requiring careful monitoring by state regulators [10].

Other pilots focus on routine care tasks that are currently bottlenecked by administrative delays. August AI has been approved to pilot AI-assisted prescription renewals for chronic conditions such as diabetes and cardiovascular disease [7]. The system reviews renewal requests and makes recommendations, which are then authorized by a Utah-licensed physician [10]. This pilot aims to prevent patients from running out of essential medications due to administrative queues. Expect Fitness is testing AI-supported telehealth for pelvic floor physical therapy, targeting women experiencing postpartum recovery issues or urinary incontinence [2]. These pilots demonstrate the state’s interest in automating low-risk, high-volume clinical decisions to improve efficiency and access.

Compass Predictive Analytics

Signal gauge

94%

Evidence Freshness

Evidence Freshness Is 94 For The Selected Signal. · Positive

tracked

Separates current evidence from aging context using a declared decay window.

9 evidence references

Signal gauge

100%

Independent Source Breadth

Independent Source Breadth Is 100 For The Selected Signal. · Positive

tracked

Shows how many genuinely independent owners support the evidence after syndication collapse.

9 evidence references

Utah regulators have authorized AI pilots that allow patients to receive initial prescriptions for conditions like acne and chronic care renewals without prior human review in later stages, aiming to bypass long dermatologist wait times. — Illustration: Utah regulators have authorized an AI pilot that can issue initial acne prescriptions without prior human review in its later stages, alongside pilots for physician-reviewed prescription renewals and pelvic floor therapy.

Oversight Mechanisms and Independent Evaluation

To address criticisms of previous pilots, Utah has implemented a more robust oversight framework. The OAIP requires third-party audits of company claims, selecting independent evaluators to stress-test system guardrails and inspect decision logs [1]. Vega Health will provide infrastructure to safely test AI model performance before deployment, benchmarking models against historical or real-time patient encounters [6]. These evaluators are tasked with validating technical claims with real-world performance data and auditing the accuracy of AI decisions [9].

The oversight architecture includes specific performance requirements that companies must meet to advance through pilot stages. For Nolla Health, advancing beyond the initial supervised phase requires targets such as at least 95% agreement between the AI and reviewing physicians, zero missed hard stops in severity screening, and zero serious adverse events [8]. Any serious adverse event triggers an immediate review and pause of the pilot. This stage-gated approach ensures that AI systems prove their safety and efficacy before gaining greater autonomy.

Despite these measures, questions remain about transparency and public accountability. The reviewed announcements do not specify how Utah will publish model error rates, adverse events, or demographic disparities [10]. Without detailed public reporting, independent observers cannot fully assess the performance claims made by companies. The state has indicated that quarterly reports will be made public and de-identified data will go to third-party auditors for external validation [8]. However, the reviewed announcements do not explain how or when Utah will publicly report on results [10].

Compass Predictive Analytics

Analytic module

module

Signal Pressure Matrix

Validated independent claim-owner cells resolve to 9 support and 0 risk pressure.

9 evidence references

Analytic module

module

Evidence Density

9 source links, 9 exact spans, and 9 independent owners support this signal.

18 evidence references

Illustration: Utah’s new oversight framework requires independent third-party evaluators to audit AI performance data before clinical deployment.

Analysis of Strategic Implications and Future Trajectory

Utah’s approach represents a deliberate strategy to position the state as a leader in AI healthcare innovation. By creating a pathway for companies to test products under regulatory relief, Utah aims to generate evidence that can inform both state and federal policy [3]. Dr. Nirav Shah of Stanford University has described Utah’s legislative laboratory as the envy of the nation, saying it is 'the only state where a company can say, "This is how we’re going to do it safely, responsibly and use this framework to see if it actually works"' [4]. This positioning allows Utah to influence national standards for clinical AI.

The expansion also highlights the tension between innovation and patient safety. While proponents argue that AI can lower costs and improve access, critics point to the risks of automated clinical decisions. The previous Doctronic pilot faced pushback from doctors who raised concerns about outdated medication risks and lack of clinical judgment [2]. Utah regulators have refused to pause such pilots, arguing that clinicians supervise AI systems during initial stages [2]. However, the move toward unsupervised prescribing in later stages requires rigorous validation to ensure patient safety.

The involvement of major health systems like Intermountain Health and University of Utah Health adds credibility to these efforts. The master agreements cover data privacy and AI governance [3], but liability remains a live question: Stanford's Michelle Mello argued that the earlier Doctronic contract is murky about who bears responsibility for potential AI-based prescribing errors [2]. As the pilots progress, the state will need to clarify liability frameworks and ensure that independent evaluators maintain true autonomy in their assessments.

The success of these pilots will depend on the ability of Utah to balance rapid innovation with robust oversight. If the state publishes comprehensive outcome data and error rates, it could establish a credible template for supervised clinical AI [10]. If transparency is lacking, the program risks becoming a cautionary tale despite its ambitious goals. The coming months will reveal whether Utah’s sandbox can deliver on its promise of improving healthcare access while maintaining high standards of safety and accountability.

Compass Predictive Analytics

Analytic module

module

Cross Pressure

Support and risk pressure differ by 100 points.

9 evidence references

Utah regulators are testing AI tools that can issue prescriptions without prior human review, marking a shift toward autonomous clinical decision-making in dermatology and chronic care. — Illustration: Utah regulators are testing an AI tool that can issue initial acne prescriptions without prior human review in later stages, while chronic-care renewals remain physician-authorized.

Conclusion

Utah’s expansion of its AI regulatory sandbox marks a decisive step in the integration of artificial intelligence into clinical practice. By approving pilots for initial acne prescriptions, chronic care renewals, and women’s health therapies, the state is testing the limits of autonomous medical decision-making [2]. The partnership with major health systems and the inclusion of independent third-party evaluators like Vega Health provide a structural framework that distinguishes this initiative from previous experiments [6]. However, the transition to unsupervised prescribing and the lack of detailed public reporting on performance metrics remain critical challenges [10].

The judgment on this initiative cannot be absolute at this stage. The pilots have not yet launched in their later stages, and the long-term safety and efficacy data are still pending [2]. Utah has demonstrated a willingness to innovate aggressively, but the true measure of success will depend on transparency and rigorous oversight. If the state publishes comprehensive outcome data and maintains strict guardrails, it may establish a model for national adoption. If it fails to provide sufficient public accountability, the program risks undermining trust in AI healthcare tools. The coming year will determine whether Utah’s sandbox delivers on its promise or becomes an example of regulatory overreach without adequate safeguards.

Utah Lets AI Examine Patients and Prescribe Without a Doctor

Expansion of the Regulatory Sandbox and Provider Partnerships

Clinical Pilots: From Acne to Chronic Care