Utah has become the first US state to authorise an artificial intelligence (AI) system to issue prescriptions for certain medications, marking a significant development in the use of AI in healthcare.The pilot programme allows Nolla Health, a healthcare technology startup, to use its AI-powered dermatology app, Nolla Derm, to examine adults with mild-to-moderate acne and recommend prescription treatments without requiring an doctor's and supervision.The programme is currently limited to adults aged 18 and above in Utah and covers a small list of topical acne medications. However, the initiative is significant as AI systems can eventually issue prescriptions without a doctors reviewing every case beforehand.The development has also raised questions about whether AI can make decisions that are normally made by licensed medical professionals, especially when a prescription depends on assessing a patient's symptoms, medical history and treatment risks.How Does Utah's AI Prescription Work?Patients using Nolla Derm are required to verify their identity, provide informed consent and complete a medical questionnaire. They then submit photographs of their face through the app.The AI system analyses the images to check acne severity and combines this information with the patient's medical history to determine whether treatment is appropriate.It can then select from a limited list of topical acne treatments approved by physicians and generate a prescription.The programme is being introduced in three stages, with different levels of medical supervision.First 100 patients: Two Utah-licensed physicians independently review and approve every AI-generated prescription before it is sent to a pharmacy.Next 400 patients: The AI can issue prescriptions directly and physicians will review cases retrospectively.Subsequent patients: Physicians review at least 10% of prescriptions each month, along with cases involving complications or side effects.Advancing in stages requires the company to meet safety targets, including at least 95% agreement with physician assessments and no serious adverse events, as well as written approval from the state. Patients can also contact a licensed physician through the app.Also read: Woman With Epilepsy Dies After Following ChatGPT’s Advice To Stop Medication: Why AI Cannot Replace A DoctorPatient Safety ConcernsWhile AI can analyse information quickly, prescribing medication involves more than identifying a condition and selecting a drug. Doctors consider a patient's medical history, allergies, existing medications, and other health conditions. There are several concerns surrounding the use of AI for these decisions.1. Incorrect diagnosisAn AI system assessing acne through photographs may not always be able to distinguish it from other skin conditions. Some skin disorders require different treatments or deeper medical investigation. If the system incorrectly classifies a condition as acne, a patient could receive inappropriate medication, which could delay correct diagnosis.Also read: New AI Tool Uses Facial Photos To Predict Cancer Patients’ Survival Risk2. Missing important medical detailsA prescription may be unsuitable because of a patient's allergies, medical history, pregnancy status or other medications.Although Nolla Derm collects medical information through a questionnaire, questions remain about whether an automated system can consistently identify important details.Also read: AI-Created Virus Injected Into A Human For The First Time To Fight Antibiotic Resistance3. Delayed treatment Not every skin problem is acne. Some lesions or changes in the skin may require an examination to rule out other conditions, including skin cancer.If patients rely on an automated assessment instead of seeking medical attention, there is a risk that a condition could be overlooked and its diagnosis could be delayed. 4. Limited ability to detect complicationsPatients may experience side effects after starting treatment. An AI system may not be able to assess these developments as comprehensively as a clinician who can examine the patient, order investigations or modify treatment.5. Accountability when something goes wrongAnother concern is determining who is responsible if an AI-generated prescription causes harm. The Utah pilot includes oversight and malpractice insurance requirements, but the bigger question of accountability remains an important issue.Doctors Question Whether AI Should Prescribe Without Human OversightAlso read: AI May Suggest 3 Times More Expensive Medical Tests And Drugs Than Doctors, New Study FindsThe initiative has prompted concerns among medical professionals about how much responsibility AI systems should be given in healthcare. John Whyte, chief executive of the American Medical Association (AMA), has questioned whether AI should independently make medical decisions."These tools can play a very effective role, but they shouldn't autonomously be trying to practice medicine, and if they are, then they need to be held to very high standards, just as humans are," he told the Los Angeles Times. "And I don't think they can meet that burden," he added.Earlier in 2026, Utah authorised another AI pilot involving prescription refills through the healthcare startup Doctronic. The programme attracted criticism from the state's Medical Licensing Board, which raised concerns about allowing AI to handle medication decisions without medical professionals.Can AI Prescribing Improve Access To Healthcare? Supporters argue that AI could help address gaps in access to medical care, especially for conditions that are commonly straightforward.Zach Boyd, director of Utah's Office of Artificial Intelligence Policy, said the programme was designed to encourage healthcare innovation while maintaining safeguards."The Nolla Health pilot program represents a significant step forward in leveraging AI to improve healthcare access for Utahns, while maintaining rigorous oversight and safety," he said. "Our office facilitates innovative AI solutions while ensuring that public health and patient safety remain paramount."Nolla's chief medical adviser, Dr Zaid Fadul, also highlighted the importance of independent oversight."The real test for AI in healthcare isn't how smart it is. It's whether someone other than the company can shut it off," he said."In Utah, the state can, and physicians can. That's the standard every AI in medicine should meet."