Even if it's archaic it's still offensive and people will use it to abuse. The problem is that YT isn't able to determine context, which is surely a much harder problem than word match. I imagine.

YT does any number of things in multiple languages. Offensive words in Burmese or any other language can easily be implemented with just a word list, which shouldn't be that hard to derive, for example you could scan a dictionary that uses the 'offensive' tag on words like English dictionaries do.