Although AGN do not typically dominate the bolometric emission of dusty star forming galaxies, large AGN fractions (sometimes > 40%) have been observed in various sub-millimeter surveys. These diagnostics have been however mostly based on X-ray counterpart selections and a complete multiwavength census of the fraction of AGN hosts is needed. I will present new advances in the modelling of panchromatic spectral energy distributions (SEDs) of active galactic nuclei (AGN), based on our publicly available code AGNfitter (Calistro-Rivera et al.2016). AGNfitter implements a fully Bayesian Markov Chain Monte Carlo method to fit the spectral energy distributions of AGNs pushing the wavelengths frontiers from the radio to the X-rays. I will present a recent application of AGNfitter on dusty star forming galaxies in the ALESS submillimeter survey to obtain an unbiased multiwavelength characterisation of the nuclear activity buried in dusty star formation. Our method reveals a significantly larger contribution of AGN activity to the emission in these galaxies than previously observed based on X-rays diagnostics. Our method represents a unique tool to potentially characterise an unbiased accretion history of the Universe when applied to larger populations of star-forming galaxies.