Fauxshield: An Explainable Deepfake Detection and Awareness Platform Using Hybrid Deep Learning
Year: 2026
Type: conference-paper
Abstract: Deepfake technology is a risk for misinformation, privacy and security. The platform we propose uses Deep Learning and Explainable AI to find deepfakes in images and videos. It has an user-friendly design. The system tries to solve the AI “black-box” problem by using LIME to explain how it works. It also includes resources to help people understand more. Our platform is built with PHP Laravel and TensorFlow. It has an API that allows other applications to use it. The goal is to make deepfake analysis easy to use for everyone: journalists, researchers organizations and regular people. The Xception model for images is expected to be 92% accurate. The CNN+RNN model for videos is expected to be 85% accurate. The overall accuracy is expected to be 88%. Our platform, FauxShield is different, from detection systems. It helps users understand deepfakes and their impact. We provide explanations of how we make our decisions. We offer support in many languages. This way we want to help people identify deepfakes and know what they mean.
Language: English
Subfield: Computer Vision and Pattern Recognition
Field: Computer Science
Domain: Physical Sciences
Open Access status: hybrid
OpenAlex ID: W7204262889