We present the task of differential face morph attack detection using a\nconditional generative network (cGAN). To determine whether a face image in an\nidentification document, such as a passport, is morphed or not, we propose an\nalgorithm that learns to implicitly disentangle identities from the morphed\nimage conditioned on the trusted reference image using the cGAN. Furthermore,\nthe proposed method can also recover some underlying information about the\nsecond subject used in generating the morph. We performed experiments on AMSL\nface morph, MorGAN, and EMorGAN datasets to demonstrate the effectiveness of\nthe proposed method. We also conducted cross-dataset and cross-attack detection\nexperiments. We obtained promising results of 3% BPCER @ 10% APCER on\nintra-dataset evaluation, which is comparable to existing methods; and 4.6%\nBPCER @ 10% APCER on cross-dataset evaluation, which outperforms\nstate-of-the-art methods by at least 13.9%.\n