Somewhere. Usually what happens is that it's drawn at the place that makes everyone unhappy. It's far enough in that some will cry censorship make casual dystopian insinuations, but well short of anything that actually makes in impact leading others to say it's ineffective because they can see obvious hate being allowed. And it happens because line-drawers are terrified of type 1 errors thanks to the media shitstorm that happens whenever it happens to someone with an audience.

This pattern happens everywhere: It's the same story with Google and obvious-spam/scams, and Amazon with fakes. Until a company has the balls to say that you're gonna have to suck it up and deal with false positives the cycle will continue forever.