Copyright © 2026 Authors retain the copyright of this article. This article is an open access article distributed under the Creative Commons Attribution License which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

@article{196318,
        author = {Sayuri Wankhede and Prof. Prachi A. Bainalwar and Snehit Telrandhe and Pratham Channujwar and Pratik Hekad and Lavannya Rangari and Vedant Kshirsagar},
        title = {Block-Chain Security Enhancement: Using Deep Learning to Detect Smart Contract Vulnerabilities},
        journal = {International Journal of Innovative Research in Technology},
        year = {2026},
        volume = {12},
        number = {11},
        pages = {3022-3046},
        issn = {2349-6002},
        url = {https://ijirt.org/article?manuscript=196318},
        abstract = {Blockchain technology has revolutionized the digital ecosystems by introducing a decentralized, transparent, and tamper-resistant systems that can eliminate the need for intermediaries among its various implementations, Ethereum is one of the leading platforms supporting the smart contracts which are the self-executing program that can automatically enforce the agreements on the blockchain. However, the rapid adoption of smart contracts has also exposed them to serious security threats. One of the most serious vulnerabilities is reentrancy. Reentrancy remains one of the most critical and exploited vulnerabilities in Ethereum smart contracts, enabling attackers to repeatedly invoke external calls before state updates and drain funds, as seen in DAO attack. Traditional detection methods- such as static analysis, symbolic execution and fuzz technique- are limited by expert-defined rules and struggle to uncover novel or evolving attack patterns. To address these shortcomings, this project proposes an automated deep-learning framework dedicated to reentrancy vulnerability detection. Solidity Smart Contracts are converted into the Abstract Syntax Tree (AST) and a graph-based representation. This process helps to capture all the present syntactic and semantic dependencies. After all these the structured features are then embedded into the neural network models, that is the Recurrent neural networks (RNNs) and another one is the Graph neural networks (GNNs), for a good precise vulnerability classification and the detection. The proposed approach combines both the static and the dynamic analysis which helps to improve the recall and also low-down the false positives which enables the robust detection of all the complex reentrancy patterns. At the end a proper user-friendly API and a web interface is being developed which provides the real-time contract scanning and detailed security reports. By focusing on the reentrancy attack detection, the system that has been proposed offers a scalable, accurate, and an automated security auditing that tries to significantly strengthen the reliability of the Ethereum-based decentralized applications.},
        keywords = {Block-Chain, Ethereum, Smart Contacts, Deep Learning, Vulnerability Detection, Solidity Code},
        month = {April},
        }