I think the issue of AI getting rights will become important well before it approaches human level consciousness. We tend to assign rights to animals (and I think rightfully so) at various levels of intelligence that are orders of magnitude below us. Most people would't think twice about stepping on a cockroach that invaded their bedroom, but if a squirrel got in they would probably go to great lengths to catch it without harming it and put it back outside. From what I've read, current general AI is around ant level intelligence, so we've got some time before it gets to the level of something like even a mouse. But I could see us using our treatment of comparable animals to define how to treat the various levels of AI.
RE: Does Artificial Intelligence Merit Moral Concern and Rights?