Deepfake Forensics Tool: AI-Driven Multi-Modal Media Authentication with Forensic Evidence Reporting

With the rapid proliferation of synthetic media generation technologies, the need for automated, accurate, and legally defensible deepfake detection has become a critical challenge in digital forensics. This paper presents a Deepfake Forensics Tool—an AI-driven, multi-modal analysis platform designed to detect manipulated media files including images, videos, and audio. The proposed system employs a fusion-based detection architecture that integrates convolutional neural networks (CNNs), frequency-domain analysis, biological signal detection, and audio spectrogram analysis to identify signs of AI-generated or manipulated content. Each uploaded media file is assigned a cryptographic SHA-256 hash and a unique Evidence ID to maintain chain of custody integrity. Experimental evaluation on benchmark deepfake datasets demonstrates that the system achieves high detection accuracy with explainable AI outputs, including heatmap visualizations and frame-by-frame analysis. The tool generates comprehensive, legally defensible forensic PDF reports containing methodology, confidence scores, heatmaps, and auditable metadata. The results confirm that multi-modal fusion analysis significantly outperforms single-model detection approaches, making the system well-suited for deployment in digital forensics, journalism verification, and legal proceedings.