Deepfakes—What Can Really Be Done Today?

Software Engineering Institute (SEI) Webcast Series

Rotem Guttman and Zach Kurtz explain what deepfakes are, how they work, and what kind of content it’s possible to create with current techniques and technology. The term “deepfake” refers to the use of machine learning to produce content for essays or to modify photos and videos. When it comes to photos and videos, the images are often so realistic that viewers are not able to tell that they are fake. In this Cyber Talk episode, Rotem Guttman and Zach Kurtz explain the kinds of machine learning that people use to create deepfakes, how they work, and what kind of content it’s possible to produce with current technology. Rotem and Zach also cover the techniques people use to create fraudulent content. Such techniques include using an actor to film a video and then replacing the actor’s face with someone else’s, as well as more advanced methods that can reproduce a person’s body movements, voice, speech, and facial expressions to make that person appear to say or do something that he or she did not actually say or do. Finally, they discuss the current limitations of these technologies and techniques, and they forecast advances that might occur in the coming years.