It's content generated by users, and it's the source of the content that matters, not the source of the moderation.
There's no generally accepted interpretation of 230 that restricts site owners from moderating user comments. https://www.eff.org/issues/cda230 "Section 230 allows for web operators, large and small, to moderate user speech and content as they see fit." See again the FB/Twitter examples I linked - those sites are also full of content generated for them, on them.
Here's a couple cases drawing the difference:
https://www.eff.org/issues/cda230/cases/anthony-v-yahoo-inc - Yahoo isn't protected against fraud claims by 230 for creating fake dating profiles, because they did it themselves
BUT
https://www.eff.org/issues/cda230/cases/universal-communicat... - Lycos is protected against defamation claims on their own message boards