The emergence of deepfake technology presents a severe threat to truth and trust in the modern digital age, when fact and fiction regularly intersect. While fighting deepfakes that target celebrities has received a lot of attention, the issue of these digital lies becoming more prevalent in the lives of regular people is growing. Our solution offers a fresh method by combining flexible learning with the robust Meso4 architecture. Unlike earlier models trained on generic datasets, our approach enables users to personalize their experience by recognizing their own unique facial attributes by building a dataset of the user’s own face that comprises both genuine and deepfake images of the user. Still, we go above and beyond simple detection. We do a thorough analysis on each video frame, looking closely for minute changes. This study highlights the urgent necessity for “democratizing security” in the digital era of video manipulation in addition to revealing a ground-breaking defense against the dangers of deepfakes. The distinctive addition of our project is Transition Point Detection, which enhances the robustness of our detection approach by detecting critical places in films when actual material changes to deepfake content.