DeepFakes: The Next Challenge in Fake News Detection

Abstract

A deepfake is a hyper-realistic video, digitally manipulated to represent people saying or doing things that never really happened. With the sophistication of techniques for developing these counterfeits, it is becoming increasingly difficult to detect whether public appearances or statements by influential people respond to parameters of reality or, on the contrary, are the result of fictitious representations. These synthetic documents, generated by computerized techniques based on Artificial Intelligence (AI), pose serious threats to privacy, in a new scenario in which the risks derived from identity theft are increasing. This study aims to advance the state of the art through the analysis of academic news and through an exhaustive literature review, seeking answers to the following questions, which we understand to be of general interest, from both an economic and a social perspective and in various areas of research. What are deepfakes? Who produces them and what technology supports them? What opportunities do they present? What risks are associated with them? What methods exist to combat them? And framing the study in terms of information theory: is this a revolution or an evolution of fake news? As we know, fake news influences public opinion and is effective in appealing to emotions and modifying behaviours. We can assume that these new audiovisual texts will be tremendously effective in undermining, even more if possible, the credibility of digital media, as well as accelerating the already evident exhaustion of critical thinking.