Which approach should a company implement to mitigate the risk of a large language model (LLM) being exploited through prompt engineering to execute harmful actions or leak confidential data in a conversational agent?