Untreated diabetes can damage multiple organ systems, including cardiovascular, nervous, urinary, and skeletal systems. Healthcare professionals report significant barriers to care, including medication access, difficulty facilitating behavioral change, and anxiety surrounding diagnosis. These barriers are particularly concerning given the rapid adoption of LLMs for healthcare support. This study examined accuracy, readability, formatting quality, and educational utility of diabetes risk assessments generated by major LLM platforms (ChatGPT, Gemini, and Copilot) in both English and Spanish.