My understanding is Reddit the company actually already uses openai's moderation api[1]. Not sure they expose that to anyone (even mods) outside the company. It's not an LLM I don't think though, it's just a set of classifiers to give a probability that something violates various categories of their TOS.

[1] https://platform.openai.com/docs/guides/moderation