I’m pretty sure commercial LLM companies have LLMs sanitize their training data before feeding it to baby LLMs, so it likely gets filtered out anyway