Also, the user might be using a language model to generate text not in English, so I should check if translation is needed. However, the response should still avoid the prohibited content. Even if the user is using a different language, the same principles apply—no explicit or inappropriate material.