Due to advances in machine learning, the tools for making deepfake audio and video content are becoming both more refined and more accessible at a rapid pace. These factors are leading to an increased incidence of deepfakes and, as a result, increased security risks. This talk will explain the fundamentals of deepfakes, including describing different types of deepfakes and the machine learning techniques used to create them. Further, security concerns relevant to deepfakes will be presented along with discussion of real-world incidents. Building on this foundation, we'll present current approaches for deepfake detection such as practical human detection methods and automated machine learning-based detection processes. A look at deepfake detection methods will include a summary of the current state of the art.
Understand the processes for creating deepfake audio and video files, and list different types of deepfake creation techniques.
Describe methods for detecting deepfakes, including both human achievable approaches and machine learning-enabled automated solutions.
Appreciate the security and safety risks that deepfakes pose and understand preventive actions that can be taken.