Honestly, it depends. The users that type those messages with the poor grammar may not be native English speakers, so they may not notice the grammatical errors for that reason. Remember: the Internet is global, so you will get messages from people from all across the world. And most people across the world aren’t native English speakers, so they won’t type perfect sentences like native English speakers would. Just see what the sentence is trying to say (in this case ‘to be honest, you’re the smartest pony I’ve seen. Tumblr can’t be trusted) and see if that message is mean or if it is kind. If it is kind, then I don’t see any reason why you shouldn’t trust it.