A Trust-Aware Authentication Framework for Preventing Account Takeover Attacks in E-Commerce Platforms

Abstract

The rapid expansion of e-commerce platforms has transformed the global digital economy, enabling convenient online transactions for millions of users. However, this growth has also increased exposure to cybersecurity threats, particularly Account Takeover (ATO) attacks, where malicious actors gain unauthorized access to legitimate user accounts. These attacks commonly exploit compromised credentials, phishing campaigns, and automated credential stuffing techniques, leading to financial loss, privacy breaches, and erosion of user trust in digital commerce systems. Traditional authentication mechanisms, especially password-based systems, are increasingly inadequate in defending against these evolving threats. This study proposes a Trust-Aware Authentication Framework designed to prevent account takeover attacks in e-commerce platforms by integrating artificial intelligence, behavioral analytics, and contextual risk evaluation. The proposed framework analyzes multiple authentication factors, including user behavioral patterns, device attributes, login location, and temporal activity patterns. A dynamic trust score is computed by combining behavioral similarity metrics, contextual reliability indicators, and anomaly detection outputs generated by machine learning models. Based on the computed trust score, the system employs an adaptive authentication strategy that allows seamless access for legitimate users while triggering additional verification procedures for suspicious login attempts. To evaluate the effectiveness of the proposed framework, multiple machine learning algorithms, including Isolation Forest, Random Forest, and Neural Networks, were implemented for anomaly detection. Experimental results demonstrate that the proposed framework significantly improves the detection rate of suspicious login activities compared with conventional authentication approaches such as password-based and multi-factor authentication systems. The results further indicate that integrating behavioral biometrics and contextual intelligence enhances the accuracy of identifying malicious login attempts while maintaining a user-friendly authentication process. The findings of this research highlight the potential of AI-driven authentication systems to strengthen cybersecurity in e-commerce environments. The proposed trust-aware framework provides a scalable and adaptive solution capable of mitigating account takeover attacks while preserving usability and trust in online commerce platforms.