February 10, 2026

Utah’s AI Prescription Renewal Pilot Could Inform Policy

AI AI governance, healthcare ai, healthcare compliance, medical regulation, prescription renewal, regulatory sandbox

By Jashaswi Ghosh and Bryant Godfrey (February 5, 2026)

Utah’s partnership with Doctronic represents the first state-approved pilot in the U.S. permitting an artificial intelligence system to autonomously renew certain prescription medications without direct physician involvement.[1]

The program is limited to routine renewals for patients with chronic conditions and is being conducted under Utah’s regulatory mitigation or sandbox framework administered by the Utah Department of Commerce‘s Office of Artificial Intelligence Policy.

Doctronic’s platform does not diagnose new conditions, initiate new treatment plans or prescribe medications for first-time use. Its authority is restricted to renewing existing prescriptions previously issued by a licensed clinician.

Patients must verify physical presence in Utah before accessing the system. The AI then reviews prescription history and guides the patient through structured clinical questions designed to surface contraindications, adverse effects or changes in condition.

If the renewal criteria are met, the prescription is transmitted directly to a pharmacy. If the system identifies uncertainty or risk, the request is escalated to a human physician for review.[2]

The pilot is limited to approximately 190 to 191 commonly prescribed maintenance medications. Categories excluded from the program include controlled substances, pain management drugs, ADHD medications, injectables, and medications used for short-term or acute treatment. These limitations were reviewed by the state and independent pharmacists prior to approval.[3]

The pilot will function as a test case for how regulators may draw boundaries between administrative automation and medical judgment.

Why Utah Is Supporting This Pilot

Utah’s rationale for supporting the Doctronic pilot reflects a policy assessment that routine prescription renewals represent a significant administrative bottleneck within the healthcare system.

State officials have emphasized that renewal visits often involve minimal clinical judgment yet consume physician time, increase patient costs, and contribute to medication lapses when appointments are delayed or unavailable.[4]

Medication nonadherence is widely cited by Utah officials and Doctronic as a major driver of avoidable hospitalizations and preventable healthcare spending.

Prescription renewals account for a substantial share of medication activity, and missed refills for chronic conditions can lead to downstream clinical deterioration. Utah views automation of low-risk renewals as a mechanism to reduce these lapses while preserving clinician capacity for higher-acuity care.[5]

The pilot also aligns with Utah’s broader strategy of using regulatory sandboxes to evaluate emerging technologies in controlled environments. The sandbox framework allows the state to temporarily modify or suspend certain licensure, scope-of-practice and telehealth requirements in order to generate empirical evidence before making permanent regulatory changes.

Utah officials have explicitly described this approach as experimental rather than deregulatory, with data collection intended to inform future legislative and policy decisions.[6]

This marks a first-of-its-kind initiative in which a state has chosen to formally collaborate with a healthcare AI company to test, observe and better understand how rapidly advancing AI capabilities may reshape healthcare delivery.

It represents an unprecedented policy experiment, pushing beyond clinical decision support into the far more consequential territory of AI-enabled prescribing without direct physician supervision.

As a pilot, its immediate scope may be limited, but its implications are not. The outcomes of this initiative are likely to influence how regulators, providers and the public conceptualize accountability, safety and the role of human judgment in the future architecture of healthcare delivery.

Measured Risk-Taking and Embedded Safeguards

Although the program permits autonomous AI decision-making, it is structured as a bounded experiment with multiple layers of oversight. Risk mitigation begins with scope control. The AI may only renew existing prescriptions within a preapproved formulary and cannot adjust dosages, initiate therapy or approve high-risk drug categories.[7]

Human oversight is phased rather than eliminated. For each medication class, the first 250 renewals are reviewed by licensed physicians to validate alignment between the AI’s decisions and accepted clinical standards.

Following this phase, ongoing sampling and auditing continue, and any case flagged as uncertain by the AI is automatically routed to a physician. Utah regulators receive regular reporting on system performance, denials, escalations and user complaints.[8]

Liability has also been addressed more directly than in many healthcare AI deployments.

Doctronic has secured malpractice insurance that applies the same standard of accountability to the AI system as would apply to a human physician. This approach reflects an effort to integrate the technology into existing legal responsibility frameworks rather than treating it as a purely informational tool.

Concerns Raised by Physician and Public Interest Groups

Despite these safeguards, physician organizations and consumer advocates have raised substantive concerns. The American Medical Association has warned that removing physicians from the renewal process risks missing subtle clinical signals that may emerge during routine interactions. Examples cited include early adverse drug effects or gradual disease progression that may not be captured through structured questioning alone.[9]

Critics also point to the reliance on patient self-reporting as a structural limitation. Patients seeking renewals may minimize symptoms or fail to recognize clinically relevant changes.

Physician groups argue that even brief renewal encounters can function as important checkpoints in longitudinal care, particularly for mental health and cardiovascular conditions.[10]

Public Citizen has taken a more categorical position, arguing that autonomous prescription renewals blur the distinction between licensed medical professionals and software applications.

The organization has been critical of Doctronic’s public framing of its platform as an AI doctor, asserting that such terminology has the potential to mislead patients and undermine established norms of medical accountability. Public Citizen has further argued that human oversight in practice may devolve into rubber-stamping once systems scale.[11]

FDA Nonintervention and Deference to State Oversight

A notable feature of the pilot is the U.S. Food and Drug Administration‘s decision not to intervene at this stage. Reporting indicates that the FDA has declined to comment, stating that the program falls outside its current regulatory purview.[12]

Former FDA officials have explained that prescription renewal falls within the practice of medicine, which has historically been regulated by states rather than the federal government.

While the FDA asserts authority over software marketed as medical devices for diagnosis, treatment or prevention of disease, it has at times deferred to state frameworks where conduct is explicitly authorized under state law. Medical marijuana has been cited as an analogous example of such deference.[13]

Importantly, this position does not preclude future FDA involvement. Sources note that if the agency determines that the AI is being marketed or deployed in a manner requiring federal authorization (i.e., the technology directly or indirectly is used for the diagnosis, treatment, prevention or cure of a disease or clinically affects the structure of function of the body), it may seek to bring the technology into compliance at a later stage.[14]

Potential Resistance From National and State Medical Boards

National and state medical boards are likely to approach Utah’s AI prescription renewal pilot through the lens articulated by the Federation of State Medical Boards in its policy on the incorporation of artificial intelligence into medical practice.[15]

That policy reflects a consistent theme across medical regulation: AI may be used to support clinical decision-making, but it does not itself constitute the practice of medicine, nor can it displace the accountability of licensed physicians.

The FSMB’s guidance emphasizes that clinical responsibility must remain clearly attributable to a licensed physician, even where AI tools are used extensively in care delivery.

Medical boards have historically resisted frameworks in which decision-making authority becomes diffuse or technologically abstracted, particularly where patients may reasonably perceive AI-generated outputs as medical determinations rather than informational support.

From this perspective, autonomous prescription renewal presents a threshold issue, not because automation is inherently impermissible, but because prescribing and renewing medications has traditionally been treated as a core clinical function requiring professional judgment and licensure.[16]

At the same time, the FSMB policy acknowledges that AI can appropriately be deployed to improve efficiency, consistency and access when used within well-defined guardrails.

These include transparency regarding AI use, physician awareness and oversight, clear disclosure to patients, ongoing monitoring of system performance, and mechanisms to identify and mitigate bias, error, or unintended consequences.

Importantly, the FSMB draws a distinction between AI that informs or augments clinical decisions and AI that independently determines clinical outcomes. That distinction is likely to be central to how boards evaluate programs like Utah’s.

Applying these principles, state medical boards may focus less on the novelty of AI-driven renewals and more on whether the regulatory structure preserves identifiable professional accountability.

Questions likely to arise include whether physicians retain meaningful supervisory authority, whether escalation pathways are robust and routinely exercised, and whether the system alters the applicable standard of care for prescription management.

Boards may also scrutinize whether patients are adequately informed that an AI system, rather than a clinician, is making the renewal determination, consistent with FSMB expectations around transparency and informed consent.

Thus, an evaluation of the present policy and guidance framework issued by FSMB policy suggests that medical boards are unlikely to categorically reject AI-enabled prescription renewal.

However, they are equally unlikely to endorse models that recharacterize autonomous AI decision-making as equivalent to physician judgment.

Utah’s pilot sits at the boundary of this regulatory philosophy, and its long-term acceptability to medical boards will likely turn on whether it can be credibly framed as structured administrative delegation under continuous professional and regulatory oversight, rather than as an alternative locus of medical authority.

Conclusion and Considerations Going Forward

Utah’s Doctronic pilot does not resolve the broader debate over autonomous AI in healthcare. Instead, it reframes the inquiry around narrow delegation, controlled experimentation and evidence-based regulation. The program’s significance lies less in its immediate clinical impact and more in its function as a test case for how regulators may draw boundaries between administrative automation and medical judgment.

Whether this model proves durable will depend on empirical outcomes, including safety data, patient adherence and clinician trust. It will also depend on how effectively regulators respond to early signals of risk rather than defaulting to either technological optimism or categorical prohibition.

As states confront rising healthcare costs and workforce constraints, pressure to automate routine functions will continue to grow. Utah’s experiment suggests one possible path forward, but it also underscores the need for careful governance, clear accountability and sustained human oversight as AI systems move closer to the core of medical practice.

Jashaswi Ghosh is counsel at Holon Law Partners LLP.

Bryant M. Godfrey is a partner and co-chair of the healthcare department and chair of the Food and Drug Administration practice group at Foley Hoag LLP. He previously served as senior lead regulatory counsel in the Center for Drug Evaluation and Research, Office of Medical Policy, Office of Prescription Drug Promotion, at the FDA.

The opinions expressed are those of the author(s) and do not necessarily reflect the views of their employer, its clients, or Portfolio Media Inc., or any of its or their respective affiliates. This article is for general information purposes and is not intended to be and should not be taken as legal advice.

Utah’s AI Prescription Renewal Pilot Could Inform Policy

Why Utah Is Supporting This Pilot

Measured Risk-Taking and Embedded Safeguards

Concerns Raised by Physician and Public Interest Groups

FDA Nonintervention and Deference to State Oversight

Potential Resistance From National and State Medical Boards

Conclusion and Considerations Going Forward

Related Posts

Preliminary Injunction Issued in the Sora “Cameo” Trademark Dispute: What It Signals for AI Product Strategy

The Social Lives of Bots: What Moltbook Reveals About Autonomous AI Communities

AI and the C-Suite: What January’s Conversation Made Clear