SiftSentinel - Unified Deepfake and Cyberbullying Detection and report generation using Explainable and Generative AI

The rapid proliferation of deep fakes and online harassment has intensified the need for robust digital content verification systems. SiftSentinel is an innovative, web-based platform designed to detect and analyze deep fake content and cyberbullying across multimodal data, including images, videos, audio, and text. The system integrates deep learning and natural language processing techniques utilizing Convolutional Neural Networks (CNNs) for visual media analysis, acoustic feature extraction for audio authenticity assessment, and NLP-based models for identifying offensive or harassing language. Trained on large-scale, heterogeneous datasets, SiftSentinel effectively detects manipulated or harmful content ranging from subtle misinformation to overt abuse. A key aspect of the platform is the incorporation of Explainable AI (XAI), which provides transparent, interpretable reasoning behind each detection outcome. Additionally, a generative AI component produces automated, evidence-based reports offering mitigation recommendations for users, moderators, and law enforcement agencies. By combining deep fake detection, cyberbullying identification, and explainable generative analysis within a unified interface, SiftSentinel contributes to enhancing online safety and promoting trust in digital environments.