Utah began piloting an autonomous prescription‑refill chatbot developed by Doctronic to streamline routine renewals and improve adherence. A Mindgard AI report, however, identified vulnerabilities in the system that could be exploited and raised questions about safety, verification, and oversight in real‑world deployments. The episode highlights the tensions between operational efficiency and risk management as health systems adopt generative AI. Regulators, developers and health systems are now grappling with auditability, patient authentication, and fail‑safe design; the case underscores the need for continuous post‑deployment monitoring and third‑party security assessment before broad rollouts.