I think ML could be used to flag certain content as potentially problematic (I'm hoping to experiment with this in the future), but yeah, it's definitely not going to be doing the final verdicts anytime soon.