Aiblocknet - Novel Framework for Authenticity Validation Using Blockchain and Machine Learning for Fake Image Detection

Abstract

The proliferation of fake images online poses a significant challenge. According to a 2022 RAND Corporation study, this issue is estimated to cost businesses $10 billion annually. Fake images erode trust and can have detrimental effects, as highlighted by a 2020 Pew Research Center survey revealing that 64% of Americans are concerned about the spread of disinformation. This work introduces a novel blockchain-based system for image authentication. The system leverages the secure and transparent ledger offered by Sepholia Testnet to create an immutable record of an image's authenticity. This is achieved by capturing and storing a cryptographic hash of the image, alongside relevant metadata, on the blockchain. Users can verify image originality by comparing a calculated hash with the one stored on the blockchain. This approach provides enhanced security, transparency, and decentralization compared to traditional methods. Furthermore, the potential integration of a deep learning model for image analysis offers additional benefits. This could significantly reduce the time spent verifying image authenticity, potentially by as much as 50% according to a 2019 Poynter Institute study focusing on journalists grappling with the vast volume of online content.

Keywords: Fake Image, Machine Learning (ML), Blockchain, Sepholia Testnet, Deepfake Detection