|
Vol.14, No.4, November 2025. ISSN: 2217-8309 eISSN: 2217-8333
TEM Journal
TECHNOLOGY, EDUCATION, MANAGEMENT, INFORMATICS Association for Information Communication Technology Education and Science |
Adaptive Authentication Using AI-Driven Password Behavior Clustering: Addressing the Reliability Gaps in Emerging Passkey Systems
Boumedyen Shannaq
© 2025 Boumedyen Shannaq, published by UIKTEN. This work is licensed under the Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 License. (CC BY-NC-ND 4.0)
Citation Information: TEM Journal. Volume 14, Issue 4, Pages 2935-2945, ISSN 2217-8309, DOI: 10.18421/TEM144-04, November 2025.
Received: 16 March 2025. Revised: 16 October 2025.
Abstract:
Recent studies regarding the topic of cybersecurity imply that the issue of password poisoning through AI is rapidly growing in severity, and additional systems of authentication that are more intelligent and behavior-aware are required. Multi-factor authentication (MFA) provides better security but is often more inconvenient for users, and existing authentication systems do not extensively leverage behavioral analysis of password creation and use. Although there has been a recent trend toward Passkey deployments instead of passwords, their effectiveness and widespread adoption have not been tested. The given work, in turn, is more realistic and empirically justified through behavioral password profiling, making it practical and reliable. This paper addresses the gap by exploring the hypothesis that clustering-based password profiling can effectively distinguish a legitimate user from a potential attacker. The password clustering model is a centroid-based model that assumes it enhances authentication accuracy by training on the password patterns users use. In the study, a list of passwords from multiple users was used to determine Centroid Password (CP) profiles via machine learning clustering. The proposed model proved more effective at classifying valid passwords and produced fewer false positives than traditional methods. The results confirm that password behavior can be categorized to enhance the authentication and verification of user identities. The research can be used to develop new, behavior-driven security models to improve the protection of digital identities against emerging AI threats.
Keywords – Authentication, security, password pattern, clustering, centroid analysis, Artificial Intelligence (AI). |
|
----------------------------------------------------------------------------------------------------------- ----------------------------------------------------------------------------------------------------------- |