If that's exactly what happened, the market would have already adjusted; no loan officer is going to say "I know poor people's social networks will indicate that they're even less responsible than they look on paper. I can't wait for them to click the 'login with Facebook' so I can finally act on that knowledge."
The actual effect will be more nuanced. This would be better at differentiating between, e.g. "People who can afford to regularly drink Johnny Walker Blue Label" (good credits), and "People who actually do" (less responsible).
Overall, it seems better to create rules that efficiently allocate resources, and then correct for inequality after the fact. That's better than creating defective resource-allocation rules as a way to keep things equal.
And please note that if you're right, you're arguing that right now, the rich subsidize the poor in this way. (I know, I know, that sounds extreme. But "Poor people only get these interest rates because it's hard to tell that they're deadbeats given current information" has to be true for your statement to work, here. You pretty much have to pick one: either this is factually wrong, and the banks that do this will lose money; or it's morally wrong, but actually does work, and we should prevent people from acting on useful information in order to let poorer people gently defraud them. This is analogous to plenty of previously or presently disastrous situations, in e.g. mortgage lending, health insurance, and student loans.)
