At a time when appearances can be false in our digital world, faith in video faces a serious threat from the arrival of deepfake innovation. This categorization of deepfake video results in popularization of malign information as one can easily find out the authentic video from a deepfake video. We show that advanced forgery detection techniques based on deep learning (DL)-based methods tested on Deep Fake Detection Challenge (DFDC) dataset help to fulfill the high demand for values robust in nature. The proposed work builds robust detection models capable of distinguishing between original and tampered media at a large scale using the extensively labeled and diverse DFDC dataset. The mitigation strategy aims to decrease the threats created by deep fakes, such as misinformation, damage to individuals ability to gain trust and security issues. The future work preserves the confidence, credit, and authenticity in digital media via DL algorithms to facilitate the style of efficient defenses for the evil effects of the deep fakes. The performance is calculated using evaluation metrics like accuracy, logloss, and, F1-score. By providing detailed analyses of experimental data, discussions on deep learning, comprehensive related work, and insights on the performance of the proposed model relative to other methods, this paper is clearly a very relevant contribution. Besides direct uses, our research encourages broader conversations about media literacy and ethics online.