| Abstract | Deep learning techniques and frameworks for manipulating or generating identities in multimedia, particularly visual data like images and videos, have advanced to a point where even individuals without specialized technical knowledge can create "deepfakes"and use them in real-world situations. This paper examines the potential threats posed by these advancements, offering an overview of manipulation techniques and state-of-the-art detection methods to counteract them. Additionally, it discusses the requirements, trade-offs, and limitations associated with these specific methods, providing a thorough understanding of the current landscape in deepfake technology and its detection, as well as possible trends for the future. |
|---|
