Deepfakes, a new type of syntetic media where the image of a person is replaced by someone else traits, recently made the headlines in many fraudolent episodes (e.g. financial scams, hoaxes and celebrity revenge videos).
In this talk we will analyse the impact of this powerful technology on a key use case: client identification or re-identification using biometric information (e.g. cameras on mobile devices).
Moreover, we will explore the latest techniques for automatic content creation and forged content detection using Deep Learning techniques (e.g. GANs – Generative Adversarial Neural Nets).
Finally, we will review a recent case study where we employed such techniques to perform liveness check of customers in financial applications.