Thanks to INTEGRAL and Swift the hard X-ray sky looks bright. For the first time in X-ray astronomy, there is an all-sky X-ray selected sample of AGNs which constitutes an unbiased census of AGNs in the local Universe. We have learned that contrary to the expectation, from the unified model, the fraction of absorbed objects, in the local Universe, is ~ 50%. As deep surveys have shown, the ratio of absorbed sources to the total population is a function of intrinsic luminosity. This has been interpreted as a decrease of the covering factor of the circomnuclear dust as a function of luminosity and finds a natural explanation in the framework of the clumpy torus model which is, as of today, well established.
Despite their lack, Compton-thick AGNs are still required to explain a substantial part of the Cosmic X-ray background around its peak. One of the main discoveries of INTEGRAL and Swift is to have unveiled a new population of Compton-thick AGNs. These are AGNs with an intense reflection component which seems to come from a material which covers a large solid angle around the central source. The current estimates show that these 'buried' super-massive black hole might be a relevant fraction ( ~ 25%) of the total population of AGNs, although given their low scattering efficiency their contribution to the CXB spectrum seems small. At the same time, large populations of Compton-thick objects can be recovered detecting the reprocessing, by dust, of UV nuclear emission into IR. The current estimates of their space density at large redshift indicates that Compton-thick AGNs are at least as numerous as the remainder AGN population. The number density of detected Compton-thick objects in the local Universe and in the high-redshift one are in agreement with predictions of population synthesis models showing that we are close to solve the mystery of the generation of the Cosmic X-ray Background. If this view is correct, hard X-ray (> 15 keV) focusing telescopes will detect many Compton-thick AGNs at intermediate redshifts.
The processing of INTEGRAL data, in the framework of the SIX survey, rests entirely on E. Bottacini whom is here warmly acknowledged. Most of the work presented here comes from my PhD thesis and I acknowledge all members of the gamma group at MPE as well as the members of the BAT-team at Goddard.