AI-Powered Decentralized Social Media Application using Ethereum

Abstract

Centralized traditional social media platforms are controlled by entities that monetize user data, restrict content visibility, and erode privacy. In contrast, current blockchain-based social media platforms are decentralized but inefficient in moderating content, resulting in the proliferation of misinformation, objectionable content, and security threats. This paper aims to bridge this gap by developing a decentralized social media platform that ensures user privacy, content moderation, and scalability without relying on centralized control. To achieve this, the system takes advantage of IPFS for decentralized file collection and Ethereum-compatible smart contracts for authentication and content verification. A major innovation of this platform is the use of an on-device-operated Natural Language Processing (NLP) model for material filtering and moderation at the user level, ensuring that no user data is collected or centrally processed. The platform will support safe text messaging, image sharing (public/private accounts), and an explore page for the discovery of public content, while all users give full control over their data. The expected results are a scalable, censorship-resistant and privacy-centric social media network, where users maintain their content ownership while AI ensures a safe digital environment.