Lol. If you think about it doesn't really add too much that's positive to the social experience. It also should have made noticable progress by now if it was a machine learning thing. I wonder if they base it on what people do like all the other ai stuff and the that's why it turns out bigoted in relief. Or maybe it's just broken.

And you can completely bypass it by just typing half of the insult Then typing “and” in between as a standalone msg And then the third msg has the 2nd half of the bad word. But entire sentences are getting muted because if one word ends in 2 letters lets say “as” and the new word starts with an s too lets say “strong” it will censor the whole sentence or a large par So “As strong “ has -blam!- in it, which gets censored. Its just a very cheap and very aggressive filter, thats probably been implemented because of the toxic pvp community and Ive seen some crazy texts in pve too. The things people wish upon others through whispers during pvp is really insane.