Given the recent progress in artificial intelligence over the last few years, various capabilities to manipulate multimedia to make people look and sound like other people have been developed. Though the technology has a lot of potential usages in the entertainment, education and activism industries, it has been leveraged by malicious users for illegal or harmful purposes. Apart from creating skepticism and spreading false information, it also pose threats to the privacy and security of individuals or groups, owing to its convincing ability to impersonate anyone. High-quality, realistic videos and images that have been digitally manipulated are referred to as ‘deepfakes’. Various researches are carried out in the field of deepfake images and certain techniques have been devised to detect the same. This paper aims to summarize the history of deepfakes, the existing generation methods, possible threats it poses in near future and why it is essential to detect the deepfake content. Further, we provide an overview of 25 research papers from 2016 to 2024 that have presented various methodologies to detect deepfake images. The analysis has been performed by grouping the techniques, to detect deepfake images, into several categories, namely, machine learning and deep-learning techniques. Moreover, various methods to detect deepfake videos have also been analyzed. Additionally, some other research questions relevant to the context have also been answered and the performance of each of the methods have been compared to conclude that deep learning methods outperform other techniques for deepfake detection.