Bit of a non-article here. There's no actual proposal on how this could be done without introducing a delay (though it could be less than radio/TV's 7 seconds) in speech messages. But delaying messages would be pretty annoying for gamers actually trying to have a conversation or coordinate actions in fast-paced games. You can't even really use prediction very effectively:
"You're a ni-" could end with "-ce person", "-fty caster", or something much worse and there's no way to tell until the person says it.
Any attempt to censor free speech(even in this context) should be met with extreme prejudice. It's a very slippery slope.
This also brings up the privacy implications.If they can detect when you swear then it's very possible they can detect certain other words or even transcribe everything. You should not expect privacy when talking on xbox live but the idea of cataloguing everything you say forever should give you goosebumps. They can use it for any purpose, trust me I know.
It seems like a noble argument until you consider that the speech you'd be protecting is people calling others, often minors, various slurs. If you doubt this play a few matches of COD.
There's the ego we must watch out for in the tech community. These people are dangerous cause they believe they know better than everyone and should be allowed to police them. It's not your place to tell people what they can and cannot say.
Though Microsoft has the right not to provide a platform for such things as a private company who doesn't live off the coffers of the tax payer. The line gets a bit blurry though as these platform become ubiquitous and becomes the main means of communication. It would be like censoring people talking over the telephone. It's a grey issue.
"You're a ni-" could end with "-ce person", "-fty caster", or something much worse and there's no way to tell until the person says it.