The advancement of deepfake technology, propelled by developments in artificial intelligence and Generative Adversarial Networks (GANs), has led to the creation of highly realistic yet synthetic media. This presents significant challenges in various areas such as cybersecurity, media authenticity, and public trust. As AI-generated content—such as videos, images, and audio, becomes harder to distinguish from real media, the need for effective detection methods becomes increasingly urgent. This systematic review examines the latest methods for detecting deepfakes, classifying them into image-based, video-based, and audio-based techniques. It provides a detailed analysis of key datasets used in training and evaluating deepfake detection systems, including FaceForensics++, DFDC, and Celeb-DF. The review also discusses major challenges in the field, such as the difficulty in generalizing detection models across different deepfake creation techniques, issues with dataset bias, and the technical challenges of real-time detection. Furthermore, the review addresses ethical and privacy concerns related to deepfake detection technologies, especially regarding their potential use in surveillance and their impact on individual rights. The paper concludes with suggestions for future research aimed at improving the robustness, accuracy, and transparency of detection models, and adapting to the evolving nature of deepfake creation methods.