Just because you can't see the dark matter doesn't mean you can't take a picture of it. Via Universe Today, here's a press release from Johns Hopkins announcing a beautiful new image of the reconstructed dark matter density in cluster CL 0152-1357 by Jee et al. (I couldn't find the paper online, but you can get a higher-resolution version of the picture at Myungkook Jee's home page.) The dark matter is in purple, the galaxies are in yellow. How do you do that? It's not because we've detected some form of light coming from the dark matter. Rather, we've detected (once again) its gravitational field -- this time, via the tiny distortions in the shapes and positions of background galaxies (weak lensing). This is a form of gravitational lensing that is so subtle you could never detect it happening to a single galaxy -- it would be impossible to distinguish between lensing and the intrinsic shape of the galaxy. But if you have a large number of background galaxies (which the universe is kind enough to provide us), you can use statistics to reconstruct the gravitational field through which the light travels, and hence figure out where the dark matter must be. Of course we're still trying to detect the dark matter, both directly (in ground-based experiments) and indirectly (looking for high-energy radiation produced by annihilating dark matter particles), not to mention using particle accelerators to actually produce candidate dark matter particles. Over the next ten or twenty years, probing the properties of dark matter is going to be one of the top priorities at the particle/astrophysics interface.
