The first one is that we assume this would be a computer simulation, i.e. a very complex program with potential bugs and such. While I'm not against the idea that some sort of computer might be behind it, the chance that it wouldn't behave anything like our computers seem more than likely. It also assumes that every projection (for lack of a better word) is the result of complex calculations, rather than the result of how the many inherent properties of a system behave together. This is something that strikes me as odd with a lot of people in sciences (broadly speaking), is that they often start assuming the tool they're using is in fact the basis of what they're trying to study with it.
My second issue, which sort of ties into the first, is that a lot of people often assume that the laws of physics would therefore be bendable. I guess the Matrix (a film I really liked, and still do) might have been partly responsible for that. It seems to me that even if we were simulated, it would in fact not change much, if anything, we would still abide by the simulated laws of simulated physics, within a simulated environment which imposes a set of restrictions on us.
Finally (sorry if this is dragging), my third issue, is a God complex one. We assume that our Simulators actually realize what they have, or rather all the intricacies of what they have made. What seems to be billions of years for us, in my view, could very well be a fraction of a second for them. Our existence may very well be (not that it currently isn't mind you) completely insignificant in their whole experiment, or whatever the simulation may be. This reminds me of a discussion with my AI professor when I was at uni, I argued that we might not be able to tell we've created a successful AI if its lifespan was too short for us to witness (I also think the first real AIs will be suicidal but that's a completely different story). It could also be that we are, in fact, unable to recognize its existence as such based on our standards.
Not really tied to any of my points, but there's also this story about simulated reality which I really like http://qntm.org/responsibility