Because there are many, many langauges and idioms in the world, and AI isn't good enough to do this accurately at the moment.

Given that Bangladesh is pretty poor and small, the likelihood is that if they were to pour resources into it, eventually the business would pull out of the country, as it wouldn't be profitable.

Note that if they block to change hands, then people will start using another idiom, so a mere filter isn't going to accomplish much here.

In Poland people use wording such as "won't sell" or "won't buy" for selling and buying stuff on facebook that cannot be traded there legally.

Bangladesh may be poor but it certainly isn't small. It's the eighth most populous country!

Apologies, I (unconsciously, I guess) weighted by advertising revenue in my head.

I should probably stop doing that.

It's certainly a cat-and-mouse-game until you've added the obvious idioms, but that only lasts a few iterations, because the users will quickly have to come up with new ones that get less useful as their understanding gets less wide-spread. It's one of those situations where you can absolutely win by and large, even if you don't catch everything. If you catch 95-99%, you're done.

Bangladesh has a population of 160m, and almost all speak Bengali, which is also used in India by a significant minority. It's not like we're talking about ignoring Liechtenstein. It won't eat up the profits by having someone add 10 rules to a filter list for a target audience of 200m+, even if they are poor.