Evaluation of the effectiveness of AI-based attacks on fingerprint biometric authentication systems
DOI:
https://doi.org/10.30837/rt.2025.4.223.08Keywords:
biometrics, authentication, fingerprint, attack, reconstruction, neural networkAbstract
The paper considers the urgent problem of information security in the context of the rapid development of artificial intelligence technologies and their use to bypass fingerprint biometric authentication systems. Сomparative analysis of modern attack methods was conducted, particularly synthetic generation technologies such as SpoofGAN and FingerFaker, as well as algorithms for reconstructing fingerprints from minutiae templates, such as the methods of Feng and Jain and also Buzaglo and Keller. The vulnerability of fingerprint authentication systems to cyber threats utilizing Generative Adversarial Networks was analyzed. It is noted that the greatest threat to personalized security is posed by reconstruction methods, which allow attackers to recover the unique papillary pattern of a specific individual from a compromised template.
Reconstruction attack method based on the Buzaglo and Keller approach was implemented. The architecture of the implemented method was described, which includes an algorithm for minutiae map formation, a ResNet50-based encoder for creating a latent vector, and a StyleGAN2 generator. Hardware and software requirements for deploying such an attack using CUDA and Python was described. The process of converting a discrete set of minutiae into a structured spatial representation and the subsequent generation of a realistic fingerprint image was described. A study was conducted using the standardized BOZORTH3 comparison algorithm to quantitatively assess the effectiveness of the proposed attack. Experimental results demonstrated the vulnerability of biometric systems to reconstruction-based attacks. It was established that even with strict security system settings with a False Acceptance Rate of 0.01%, the reconstructed fingerprints achieved a recognition accuracy of 96.68%.
A comparative analysis of the effectiveness of the implemented method against other AI-based attacks was performed. It was determined that while synthetic generation methods demonstrate high overall accuracy, the reconstruction method is critically dangerous specifically for targeted attacks on specific accounts.
References
Методи і технології біометричної ідентифікації за результатами літературних джерел / Л. Г. Коваль та ін. // Вчені записки ТНУ ім. В.І. Вернадського. Серія: технічні науки. 2019. Т. 30 (69) Ч. 1 № 2. С. 104–111.
Security vulnerabilities against fingerprint biometric systems: book chapter / Sushanta Kumar Das and other // Recent Advances in Computational Intelligence Applications for Biometrics and Biomedical Devices. 2025. Chapter 4. Р. 35–43.
Generating MasterPrints for Dictionary Attacks via Latent Variable Evolution / P. Bontrager, et al. // Conference: 2018 IEEE 9th International Conference on Biometrics Theory, Applications and Systems (BTAS) IEEE BTAS. 2018. P. 1–9. DOI:10.1109/BTAS.2018.8698539.
Goodfellow I, Pouge.t-Abadie J, Mirza M, Xu B, Warde-FarleyD, Ozair S, Courville A, Bengio Y. Generative adversaria lnetworks. Commun // ACM, 2020. vol. 63, no. 11. Р.139–144.
Ateniese G. Hacking smart machines with smarter ones: How to extract meaningful data from machine learning classifiers / G.Ateniese,G.Felici,L.V.Mancini 1 et al. // Іnternational Journal of Security and Networks. 2010. Vol.10,Issue 3. P.137–150.
Combined Adversarial Attacks for Breaking Fingerprint Presentation Attack Detectors / S. Zedda // EUSIPCO 2025: conference paper, Thu, 11 Sep. 2025, Italy, Bellini. Р. 1352–1356.
Yu Y, Niu Q, Li X, Xue J, Liu W, Lin D. A Review of Fingerprint Sensors: Mechanism, Characteristics, and Applications. Micromachines (Basel). 2023 Jun 14;14(6):1253. doi: 10.3390/mi14061253. PMID: 37374839; PMCID: PMC10305017..
Maltoni D., Maio D., Jain A.K., Prabhakar S. Handbook of Fingerprint Recognition. 2nd ed. London : Springer-Verlag London Limited, 2009. 475 p.
S. A. Grosz, A. K. Jain. SpoofGAN: Synthetic Fingerprint Spoof Images // IEEE Transactions on Information Forensics and Security. 2023. vol. 18. Р. 730–743. doi: 10.1109/TIFS.2022.3227762.
Shen Y., Ma Z., Lin F., Yan H., Ba Z., Lu L., Xu W., Ren K. FingerFaker: Spoofing Attack on COTS Fingerprint Recognition Without Victim's Knowledge // Proceedings of the 21st ACM Conference on Embedded Networked Sensor Systems (SenSys '23). New York : Association for Computing Machinery, 2024. Р. 167–180.
K. Cao, A. K. Jain. Learning fingerprint reconstruction: From minutiae to image // IEEE Trans. on Information Forensics and Security. 2015. Vol. 10, no. 1. Р. 104–117.
Bouzaglo R., & Keller Y. Synthesis and reconstruction of fingerprints using generative adversarial networks. arXiv preprint arXiv:2201.06164, 2022.
NIST Biometric Image Software (NBIS). nist.gov. URL: https://www.nist.gov/services-resources/software/nist-biometric-image-software-nbis/.
Downloads
Published
How to Cite
Issue
Section
License
Authors who publish with this journal agree to the following terms:
1. Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
2. Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
3. Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work (See The Effect of Open Access).
