Cluster lensing is the strongest evidence we have for the existence of dark matter. Generally, cluster lenses are well modelled as elliptical, extended halos of dark matter around the most luminous galaxies in a cluster. I will describe how the ellipticity and density profile of the dark matter distribution in a cluster offers a powerful probe to the nature of the dark matter. Cosmological simulations of pure collisionless dark matter predict that halos should have cuspy profiles, which do not fit the rotation curves of dark-matter dominated dwarf galaxies. I will show that density profiles of clusters of galaxies are also in disagreement with these predicted profiles. I will then discuss why the observed elliptical distribution of dark matter eliminates one of the popular solutions to this discrepancy: self-interacting dark matter. I will then show that if the dark matter is assumed to have the same core radius as the distribution of stars in the central cluster galaxy, there is excellent agreement with the constraints imposed by lensing observations. This suggests that the constant density cores in dark matter halos are due to baryonic effects, and are caused by the same process that produces cores in giant elliptical galaxies.