International Journal of Scientific Research and Engineering Development

The increased complexity of face-swap deepFake videos makes them more difficult to differentiate with real videos and poses significant threats of misinformation and undermined reputation. This paper introduces an AI/ML-based solution built to detect such manipulations accurately and efficiently by spotting subtle artifacts. We propose a hybrid model that integrates Convolutional Neural Networks (CNNs) and Long Short-Term Memory (LSTM) networks to capture both spatial and temporal patterns in video data. The CNN extracts spatial details from individual frames, such as visual artifacts, while the LSTM examines temporal inconsistencies between frames, like unnatural motions or expressions. This combined approach aims to deliver robust, adaptable model capable of achieving high accuracy in distinguishing authentic videos from deepfakes, helping preserve digital trust.