What's wrong with AGN models for the X-ray background ?

The origin of the hard X-ray background (XRB) as a superposition of unabsorbed and absorbed Active Galactic Nuclei is now widely accepted as the standard model. The identification of faint X-ray sources in ROSAT, ASCA, and BeppoSAX medium-deep surveys and their average spectral properties are in broad agreement with the model predictions. However, AGN models, at least in their simplified version, seem to be at odds with some of the most recent findings calling for substantial revisions. I will review the recent XRB "best fit" models and discuss how the foreseen XMM and Chandra surveys will be able to constrain the allowed parameter space.