Forensic AI: A Novel Multi-Granular Approach for Detecting Synthetic Media Manipulation

With the rapid evolution of AI-driven synthetic media, deepfake technology has emerged as a potent tool for deception, raising concerns about misinformation, security, and digital authenticity. This study introduces a novel multi-granular artifact detection framework to distinguish between real and AI-generated images with high precision. By leveraging deep learning techniques and a custom convolutional architecture, the approach identifies intrinsic and extrinsic inconsistencies left by synthetic media generation processes. Extensive experiments across diverse datasets validate the robustness of the model, demonstrating superior generalization across both seen and unseen deepfake samples. The research provides a step forward in forensic AI, ensuring enhanced media integrity and mitigating the growing threats posed by deepfake technology.