Utah’s AI Medication Prescribing Risks: Navigating the Challenges of Autonomous Systems
In this update we examine how Utah’s groundbreaking partnership with an AI company has introduced serious AI medication prescribing risks by authorizing unsupervised prescribing of nearly 200 medications.
In January 2026, Utah partnered with Doctronic to deploy the first system in the United States authorized to prescribe corticosteroids, statins, antidepressants, hormones, and anticlotting agents without any physician involvement. While promoted as a solution to medication nonadherence and primary care shortages, the initiative’s deliberate bypass of state licensing laws and federal FDA oversight has triggered significant legal, safety, and public health concerns.
Bypassing Physician Oversight for Chronic Care
The program targets prescription renewals for chronic conditions, responding to well-documented gaps in medication adherence. Nonoptimized regimens impose an annual economic burden of $528.4 billion in the United States. Proponents argue that AI prescribing could reduce errors, improve efficiency, expand access amid physician shortages, and free clinicians for complex care.
Hidden Dangers of Unregulated AI Prescribers
These potential benefits are overshadowed by substantial AI medication prescribing risks. Unlike physicians who complete rigorous training and certification, AI systems lack equivalent safeguards. Doctronic has not disclosed its core algorithms, providing only vague assurances that the system is “trained on established medical protocols.” Its sole performance claim—an 81% match on primary diagnosis—was generated by company-affiliated authors, never peer-reviewed, and studied a different clinical setting. Such gaps raise serious doubts about the technology’s readiness for unsupervised use.
State Sandbox Creates Federal Conflict
Utah is using its regulatory sandbox statute to temporarily waive laws requiring prescriptions to come from licensed practitioners. This move directly conflicts with federal law (21 USC §353), which states that prescription drugs must be dispensed pursuant to a prescription from a licensed practitioner or risk being deemed misbranded. Although several states maintain similar sandbox programs, Utah’s application to fully autonomous AI prescribing is unprecedented.
FDA’s Role sidelined in AI Device Regulation
An autonomous AI prescriber meets the legal definition of a medical device under federal law and should require FDA review because it replaces, rather than assists, physician judgment. Doctronic has avoided FDA engagement, claiming the agency does not regulate the practice of medicine. However, current political signals—including executive actions discouraging AI regulation—suggest limited federal enforcement is likely in the near term.
Impact on Health Economics and Market Access
The $528.4 billion annual cost of suboptimal medication use makes this area ripe for economic evaluation. However, the absence of independent, peer-reviewed safety data prevents credible Health Economics and Outcomes Research (HEOR) modeling. Drug manufacturers may see increased utilization but also face new liability risks, while payers may question whether AI-generated prescriptions meet reimbursement standards that require a valid clinician-patient relationship.
AI medication prescribing risks will only be mitigated through rigorous, transparent evidence rather than regulatory evasion. Physicians, health systems, and policymakers must insist that AI tools meet the same legal and clinical standards long required of human practitioners to protect patient safety while harnessing genuine efficiency gains.
