FWIW, this is a common ‘problem’ across CG apps. However, it isn’t really a problem, it was designed that way. You don’t want an anti-aliased depth pass, because it will, in fact, create more visual artifacts. It will make the edges turn darker, and your compositing software will interpret those pixels as being further in the background causing worse results than what you started with.
What you’re supposed to do is as someone mentioned earlier: double the depth pass resolution and scale it down by half. It’s essentially the equivalent of what antialiasing does, but doing it this way will keep those data values intact.