This 25-minute information session will explore the potential risks of inputting unpublished research and data into large language models (LLMs) and discuss potential mitigation strategies.