You are developing a chat application that uses Azure OpenAI GPT-4 to answer customer questions. You need to ensure the model does not generate harmful content. Which configuration should you set?