That just means the person writing the algorithm becomes the authority. As we recently saw with Amazon's hiring practices, algorithms have bias as well. Granted, a better designed system might not have made this mistake, but since humans designed it, can it ever be perfectly unbiased?

Without actual censorship (which I'm against) the offending material still exists, but would only be found by those who seek it. I do like that approach personally, but I don't think it would be enough to satisfy many people who prefer to eradicate offensive speech through authoritarian measures.