Due to the recent rapid development of AI, a variety of techniques and technologies are now employed to alter multimedia. Although we had the technology, it was frequently mishandled or exploited for nefarious or criminal ends rather than for legitimate uses like entertainment and education. Recently, the term “Deep fake” has come to refer to the realistic, high-quality videos and images that have been altered. In a literature review, numerous methods for deep fake detection have been described. It is essential to develop detection techniques that can thwart these kinds of frauds and enhance the study of video and audio forensics. With the use of techniques like Convolutional Neural Networks (CNN), Xception Network, Recurrent Neural Networks (RNN), Long Short Term Memory (LSTM), and others, we demonstrate the numerous Deep fake creation and detection approaches currently being researched. These techniques will act as the cornerstone for the creation of a new deep fake detection system that is more accurate and concise. We also performed a comparison examination of several approaches using a standard methodology.