Recently, generative AI has affected many sectors in several ways, including developments in language generation in GPT and deepfake technology. Deepfakes, which use deep learning in media content, pose significant risks to security, the political sphere, and public confidence. This review paper covers an enormous amount of information regarding deepfake and face manipulation, including attribute manipulation, image synthesis, face swap, and re-enactment. Thus, we classify deepfake detection techniques based on a novel proposed taxonomy to provide a clearer view of existing solutions. Also, we explain a comprehensive method for detecting deepfakes in images and videos, with details concerning data gathering, data pre-processing, feature extraction, and intricate AI methods. The primary goal of the proposed study is to increase understanding and generate new ideas to fight against the rapidly increasing evil of deepfakes.