A Generative AI Engineer has developed an LLM-based application to provide answers about internal company policies. The engineer needs to ensure the application avoids hallucinating information or leaking confidential data. Which method is NOT suitable for preventing hallucination or data leakage?