I mean think about it. When Twitter tries to implement the same AI they use to ban ISIS, and the system receives feedback to correct it every time a Republican who's not obviously a Nazi gets banned, the system is still unable to find a criterion for leaving many other Republicans unbanned when they join up. It can't tell the difference, they haven't figured out a mathematical algorithm that bans overt white supremacists and leaves nearly all Republican officials and supporters untouched. Same opinions, different language and nuance. Or maybe it's KGB math, I dunno.