As generative AI becomes a tool for public health communication, its visual outputs must be clear, culturally sensitive, and medically appropriate. GPT-Generated Health Visuals introduces a rubric-based framework to evaluate the effectiveness of AI-generated health imagery and examines how prompt design impacts clarity and accessibility. The project highlights communication-focused risks and opportunities in visual prompt engineering, culminating in a generalization study and practical guidance for using AI to support inclusive health education. Intern: Chloe Chung Mentor: Samantha Levy (AOS)