Survey of Techniques Available for Digital Forensics Technique Based in AI to Detect Deepfake Video and Image

The proliferation of deepfake technology has emerged as one of the most significant threats to digital media integrity in the contemporary era, necessitating robust forensic approaches capable of distinguishing authentic content from synthetically generated manipulations. This survey paper provides a comprehensive examination of artificial intelligence-based digital forensics techniques developed for deepfake video and image detection. The paper systematically reviews the evolution of detection methodologies, ranging from traditional machine learning approaches to advanced deep learning architectures, including convolutional neural networks, recurrent neural networks, generative adversarial networks, and transformer-based models. The survey explores the fundamental artifacts exploited by detection algorithms, such as spatial inconsistencies, temporal anomalies, biological signal discrepancies, and metadata irregularities. A critical analysis of existing literature reveals significant research gaps, particularly concerning cross-dataset generalization, adversarial robustness, and the detection of unseen generation methods. Based on these findings, the paper proposes a comprehensive methodology incorporating multi-modal feature extraction, ensemble learning, and explainable AI principles to enhance detection reliability. The expected outcomes include improved detection accuracy, reduced false positive rates, and increased interpretability of forensic decisions. This survey concludes by highlighting future research directions, including the development of real-time detection systems, blockchain-based content provenance mechanisms, and collaborative forensic frameworks that integrate human expertise with machine intelligence.