A healthcare startup is training a generative AI model on patient records to summarize clinical notes. The compliance team is concerned that sensitive personal information could be memorized by the model and later exposed in generated outputs. Which Google responsible AI principle most directly addresses this concern?