This article discusses the development of a synthetic PHI generation pipeline for medical imaging AI, focusing on creating privacy-safe training and validation data. It addresses a critical problem in AI development related to protecting personal health information (PHI).