The recent controversy in Utah over the Doctronic AI pilot program, as reported by STAT News, underscores a critical gap in healthcare regulation: the absence of a coherent framework for licensing autonomous AI systems as medical practitioners. Utah’s Medical Licensing Board suspended the program, which allows a chatbot to evaluate patients and recommend prescription renewals for chronic conditions, citing a lack of clinical oversight and potential risks to citizens. This incident is not an isolated failure but a symptom of a broader systemic issue. With 47 states considering over 250 bills on clinical AI, the regulatory landscape is fragmented, and the federal FDA framework—designed for static medical devices—struggles to accommodate adaptive, generative AI systems. This article delves into the implications of this gap, explores missed dimensions in the original coverage, and proposes a path forward by synthesizing multiple sources and contextual patterns.

The STAT News opinion piece correctly identifies the mismatch between current regulatory tools and AI’s dynamic nature. However, it misses the deeper historical parallel to the early days of telemedicine regulation in the 1990s and 2000s, when states also grappled with fragmented rules for remote care before federal guidelines emerged. Just as telemedicine required new licensure compacts and cross-state agreements, AI in healthcare demands a unified credentialing model. The original coverage also underplays the urgency of the physician shortage—projected to reach deficits of up to 86,000 doctors by 2036 per the Association of American Medical Colleges (AAMC)—and how AI could address disparities in rural and underserved areas if regulatory barriers are resolved. Instead, it focuses narrowly on Utah’s misstep without connecting to the broader pattern of state-level experimentation colliding with clinical governance.

Peer-reviewed research adds weight to the argument for AI’s potential. A 2025 prospective study in Kenya, published in The Lancet (n=39,800 primary care visits), found that AI-supported clinicians reduced diagnostic and treatment errors by 31% compared to human-only care (high-quality RCT, no conflicts of interest reported). Similarly, the NOHARM trial (2025, n=1,200 clinical tasks), published in NEJM, showed large language models matching or exceeding physician performance in routine primary care tasks (high-quality RCT, industry funding disclosed). These studies suggest AI’s clinical adequacy is no longer speculative but proven in controlled settings. What’s missing is a bridge from research to practice—a licensing model that ensures competency without stifling innovation.

The Utah case also reveals a governance blind spot: the lack of pre-deployment validation. As the STAT piece notes, the state phased out physician review prematurely, prioritizing scalability over safety. A better approach, inspired by human medical licensing, would require AI systems to pass standardized clinical exams (akin to the USMLE), undergo supervised deployment with real-world data, and face continuous performance monitoring. This mirrors the credentialing of nurse practitioners, who must demonstrate competency through education, exams, and supervised practice before independent work. Why should AI be held to a lower standard? Furthermore, the original coverage overlooks the risk of bias in AI training data—a known issue flagged in a 2023 systematic review in JAMA Network Open (n=45 studies, observational, no conflicts reported)—which could exacerbate health inequities if not addressed in licensing criteria.

The stakes are high. Without a federal licensing framework, state-level patchwork policies will continue to delay deployment, increase costs for developers, and limit access for patients in need. The physician shortage, coupled with AI’s proven potential, makes this a pivotal moment. A national credentialing body for AI doctors—perhaps under the Department of Health and Human Services—could standardize competency testing, bias audits, and update protocols, much like the Federation of State Medical Boards oversees human licensure. Until then, incidents like Utah’s will recur, stalling a technology that could transform healthcare delivery.