MUMBAI, India, Aug. 17 -- Intellectual Property India has published a patent application (202641095685 A) filed by Sns College Of Technology on August 07, 2026, for Real-Time Deepfake Detection And Verification Using Artificial Intelligence For Secure Digital Media Authentication.

Inventors include A. S. Arul Sowmiya; Shasikesan P; Siva Prakash S; Sijay T S; and Shiva Shamith K K.

The application for the patent was published on August 14, 2026, under issue no. 33/2026.

Abstract: This invention introduces an Artificial Intelligence (AI)-driven Real-Time Deepfake Detection and Verification System designed to improve the authenticity, reliability, and security of digital images, videos, and audio content. Conventional deepfake detection methods often struggle with increasing manipulation sophistication, delayed analysis, limited scalability, and reduced detection accuracy. The proposed system overcomes these challenges by integrating deep learning models, facial landmark analysis, temporal inconsistency detection, voice verification, and blockchain-based authentication within an intelligent verification framework. The system continuously analyzes multimedia content using Convolutional Neural Networks (CNN), Vision Transformers (ViT), Recurrent Neural Networks (RNN), and Explainable AI (XAI) techniques to identify synthetic manipulations in real time. Adaptive verification algorithms automatically classify media authenticity while maintaining high detection accuracy with minimal processing delay. Cloud connectivity enables centralized monitoring, automated reporting, continuous model updates, and secure digital certificate generation. The modular architecture allows deployment across social media platforms, government organizations, news agencies, financial institutions, law enforcement agencies, and digital forensic laboratories. By reducing misinformation, preventing identity fraud, and improving trust in digital communication, this invention supports secure and responsible digital ecosystems.

Disclaimer: Curated by HT Syndication.