I'm pretty sure they could implement some manner of detection/security software if they actually wanted to.

However, because they lack such a thing it seems to make them reluctant to ban bots that aren't RMT spammers and/or blatantly hacking in the extremely off-chance they could end up punishing who's innocent..