A Distributed Authenticity Verification Scheme Using Deep Learning for NFT Market

Abstract

With the recent proliferation of blockchains, identifying security risks to them has become an important issue. Among the various types of cyberattacks against blockchains, the blockchain poisoning attack involves the storing of malicious data in the blockchain to compromise it. One scenario is an attack that distributes forgeries of digital content traded and managed using Non-Fungible Token (NFT) on the blockchain. Currently, concomitant with the growing interest in NFT-based content trading, blockchain poisoning attacks on NFT trading and their effects have also increased. In this study, we examined the issues that may lead to attacks in the process from generation to distribution of digital content using NFT from the viewpoint of flexibility and interoperability of the content. Consequently, we discovered that there are two types of attack risks in NFT trading using malicious content: fake attacks and reuse attacks. As a countermeasure against these attacks, we propose a method for verifying the authenticity of the content itself using a decentralized scheme. The proposed method ensures the confidentiality of contents by using deep learning as an irreversible transformation operation in the distributed scheme and for privacy protection.