Software driven text censorship is really really hard to make fool proof. People are going to be upset about this and really try to beat the system which will probably make the end result even worse.
They probably should have stuck with traditional bad word censorship and had a manual review system for racist/sexist sounding things that the system flagged.
They probably should have stuck with traditional bad word censorship and had a manual review system for racist/sexist sounding things that the system flagged.
