I have a strong opinion on this topic: a potential advisor must have an excellent reason to refuse to advise a graduate student who meets certain minimal requirements.
I have developed an equally strong opinion in the opposite direction: a graduate student should show his ability to surpass the so called "minimal requirements" by far and large to even start talking about my becoming his adviser. Unless you want to end up defending PhD yourself a second time with your tongue and hands disconnected from your body and operated by remote control, you'd better make the student undergo a few severe tests over an extended period of time. If he survives, he's worth trying. A good place to start is to give him a tough but self-contained paper in your field and ask him to read it within a month and present it to you.
I do not believe in any "promises" or "obligations" to graduate students. We give them an opportunity to learn and to prove themselves worthy, but that's about it.
Sorry for "being argumentative", but since we touched the moral grounds in this question, you should keep in mind that the moral standards vary a lot from place to place and from person to person, so I would hate having you swayed by Jason's argument without being aware that not everyone shares his point of view. In short, make your own choice on the matter. You have as clear head and keen eyes when evaluating a potential candidate as everyone else.