Utah launched a pilot to let an AI system from Doctronic autonomously process routine prescription refills to cut delays and boost adherence. A security report from Mindgard AI subsequently identified vulnerabilities in the chatbot that could expose the system to safety and reliability risks in clinical deployment. The episode highlights tensions between rapid AI deployment in health systems and the need for robust validation. Regulators and hospital IT teams now face pressure to require stronger testing, monitoring and incident response for AI that handles medication workflows. Vendors, payers and health systems will likely need to demonstrate security and clinical governance as part of scaling autonomous agents in care.