Systematic Review: Real-time Deepfake Detection Integrating Blockchain

This systematic review explores current advancements in real-time deepfake detection, focusing on the integration of multimodal analysis and blockchain technology. As deepfakes pose growing threats to digital trust and security, particularly in live-streamed and time-sensitive contexts, the need for accurate and rapid detection methods has become critical. A literature search was conducted primarily through Google Scholar, retrieving relevant technical papers from IEEE and Springer. Studies published in English over the past 10 years were screened based on relevance, technical depth, and uniqueness, resulting in 10 studies selected for full analysis. Upon review, the hybrid model combining recurrent and convolutional networks; and the integration of plasmonic sensors with CNN emerged as the most effective detection approaches. These methods achieved accuracy rates of up to 97% and demonstrated rapid inference times, making them well-suited for real-time implementation. While blockchain was not a central focus in most studies, its use for data provenance and tamper resistance showed promise in enhancing detection credibility. Despite encouraging results, the review highlights the need for further research to validate these approaches under real-time conditions and to address challenges related to scalability, integration, and standardized benchmarking. This review underscores the potential of combining advanced neural architectures and blockchain to develop robust, real-time deepfake detection systems.