So long as we aren't using the AI to punish the posters. Vis-a-vis the google incident last year with the guy loosing his entire account because of telemedicine.
I'm okay with false positives. You took down my video of a sphinx cat because your AI fagged it as NSFW. fine. I can't think of a scenario where random people being able to post things getting blocked occasionally for things being incorrectly flagged is a risk. The risk comes when they then use that to ban and delete accounts. If they just lock the account in a non-destructive way, i.e. block new posting/shadow ban. and reverse it on appeal.
This is of course excluding the use of such tools in a non-good faith way.