(Do I trust AI to do a first-pass review? No, I don't.)
You can reduce that to some degree by putting a threshold of number of views. But that just moves the problem. Then you won't have many posts that exceed the threshold. (Though it reduces the problem somewhat, because social media posts that exceed the threshold will be the one that they make the most money from.)
But the worse problem is, who decides what's "damaging"? The politicians do. That means that posts that are damaging to the politicians are going to be among the first things removed. That makes this a very dangerous path.