The best practice is to always ask an LLM if your sentences sound natural before committing. For example, you definitely don't wanna call men's boxers "panties". If you described the context to the AI, it would have corrected you.
If AI says the topic is too sensitive, try Grok or local models. You'll improve very quickly and soon you won't need to consult AI anymore.