An AI arbitrarily banning someone for a completely innocuous post isn't broken?
It's a force multiplier that every social media platform needs to not drown in costs for human moderation. Especially human moderation as described above.
Unless you get people to pay for social media. Good luck with that.
If that ban by AI is available for review and appeal, it's just triage that admits false positives. The alternatives that I can fathom is to either allow unsavory and potentially harmful content to be present on your platform until human eyes get on them, or to have all comments await moderation until human eyes get on them.