Work

ChainML Guard: Predictive artificial intelligence for Ethereum fraud prevention

Year: 2026

Type: software-paper

Abstract: Even though the Ethereum ecosystem is based on decentralization and immutability, it remains exposed to phishing-related scams and malicious addresses. Because attackers can rapidly create multiple wallets for fraudulent purposes, traditional security mechanisms based on static blacklists provide limited protection against newly emerging threats. In this context, ChainML Guard introduces a proactive solution for Ethereum transaction security. Before a transaction is broadcast, the system estimates whether the sender address is malicious by using a Multi-Layer Perceptron (MLP), a deep learning feedforward neural network, trained on lightweight address-level indicators, namely wallet balance, transaction count, and wallet age. By leveraging the Etherscan API, ChainML Guard retrieves these features and transforms raw blockchain data into a real-time risk score that supports immediate transaction decision-making. The solution is deployed through a Flask-based interface, providing an adaptive protection layer for proactive sender-address screening in decentralized environments.

Source: SoftwareX

DOI: 10.1016/j.softx.2026.102955

Authors: Mahmoud Bidry, Mohamed Hanine, Hiba Anini, Ouiame Rehouni, Waeal J. Obidallah, Hisham Alhulayyil

Institutions: Chouaib Doukkali University, Imam Mohammad ibn Saud Islamic University

Language: English

Topic: Blockchain Technology Applications and Security

Subfield: Information Systems

Field: Computer Science

Domain: Physical Sciences

Open Access status: gold

Estimated APC paid: $1,200

Funders: Deanship of Scientific Research, Imam Mohammed Ibn Saud Islamic University, Imam Mohammed Ibn Saud Islamic University