Cutting-Edge Real-Time System for the Detection of AI-Generated and Manipulated Video Content

Abstract

This study provides a social media network inspired by Instagram that incorporates an integrated AI-based system for video tampering detection. For deepfake signatures, the system analyses submitted video information using MesoNet, a compact convolutional neural network. To guarantee the authenticity of the content, detected altered videos are immediately marked with a visible watermark before being shared. To create a safe, scalable multipage application, the project makes use of Node.js, MongoDB, Python, and MoviePy. Results reveal 85–90.