Deepfake Analyzer

Deepfake videos have gotten alarmingly realistic, and spotting the difference between genuine footage and clever fakes is now a critical task for everyone from cybersecurity teams to journalists.In this work, we trained and tested several machine learning models on the FaceForensics++ dataset-packed with both real clips and manipulated ones-and found that MobileNetV2 delivered the best results with 96% accuracy.Here, we share how we set up our experiments, the insights we gained, and what the road ahead looks like for making detection tools even more robust.