Data Privacy in AI: Securing the Sensitive Information Through Homomorphic Encryption

Year : 2025 | Volume : 03 | Issue : 02 | Page : 25-30
    By

    Pawar Pratik Suresh,

  • Netra Patil,

  1. Student, Master of Computer Application, Sinhgad Institute of Business Administration and Research, Pune, Maharashtra, India
  2. Director and Professor, Master of Computer Application, Sinhgad Institute of Business Administration and Research, Pune, Maharashtra, India

Abstract

document.addEventListener(‘DOMContentLoaded’,function(){frmFrontForm.scrollToID(‘frm_container_abs_198176’);});Edit Abstract & Keyword

Artificial intelligence (AI) technology increasingly relies on sensitive user data, particularly finance and healthcare. While legacy encryption technologies safeguard data in transit and at rest, they are of no use when data must be decrypted to be processed. This is a bleak privacy threat, particularly in AI applications that call for constant processing of data. The objective of this study is to apply homomorphic encryption, a feature in which operations are carried out on the encrypted data themselves without disclosing the data. Particularly, we take the CKKS (Cheon-Kim-Kim-Song) scheme into account to approximate calculation of encrypted real-number data. A logistic regression model was executed using the TenSEAL Python library in an attempt to perform encrypted inference on sensitive data. What is demonstrated in our implementation is the possibility of performing private AI computation without sacrificing the accuracy of predictions. Even as homomorphic encryption does create some computation overhead, our experience confirms that it likewise creates an actual-world performance-to-data-privacy trade-off. The method achieves raw data protection not possible during the processing horizon, thus enabling regulatory compliance as well as end-user trust. This makes the method highly worth it for applications that are sensitive to privacy such as digital health platforms, risk analysis in financials, as well as in sensitive analytics. In short, this research illustrates the feasibility of integrating homomorphic encryption into AI processing to secure confidential information while processing. It provides an entrance to privacy-focused, secure AI systems that provide data utility without exposing raw inputs.

Keywords: Training, placement, management, administrative, database

[This article belongs to International Journal of Information Security Engineering ]

How to cite this article:
Pawar Pratik Suresh, Netra Patil. Data Privacy in AI: Securing the Sensitive Information Through Homomorphic Encryption. International Journal of Information Security Engineering. 2025; 03(02):25-30.
How to cite this URL:
Pawar Pratik Suresh, Netra Patil. Data Privacy in AI: Securing the Sensitive Information Through Homomorphic Encryption. International Journal of Information Security Engineering. 2025; 03(02):25-30. Available from: https://journals.stmjournals.com/ijise/article=2025/view=0


document.addEventListener(‘DOMContentLoaded’,function(){frmFrontForm.scrollToID(‘frm_container_ref_198176’);});Edit

References

  1. Acar A, Aksu H, Uluagac AS, Conti M. A survey on homomorphic encryption schemes: Theory and implementation. ACM Comput Surv. 2018 Jul 25; 51(4): 1–35.
  2. Gentry C. A fully homomorphic encryption scheme. PhD Thesis. Stanford (CA): Stanford University; 2009; 1–190.
  3. Cheon JH, Kim A, Kim M, Song Y. Homomorphic encryption for arithmetic of approximate numbers. In International conference on the theory and application of cryptology and information security. Cham: Springer International Publishing; 2017 Nov 30; 409–437.
  4. Microsoft. (2023). Microsoft SEAL: Fast and Easy-to-Use Homomorphic Encryption Library. Microsoft Research. [Online]. Available from: https://www.microsoft.com/en-us/research/project/microsoft-seal/
  5. GitHub. (2024). OpenMined/TenSEAL: A library for doing homomorphic encryption operations on tensors. [Online]. GitHub. Available from: https://github.com/OpenMined/TenSEAL
  6. Chen N, Li J, Zhang Y, Guo Y. Efficient CP-ABE scheme with shared decryption in cloud storage. IEEE Trans Comput. 2020 Dec 14; 71(1): 175–84.
  7. Froelicher D, Troncoso-Pastoriza JR, Pyrgelis A, Sav S, Sa Sousa J, Bossuat JP, Hubaux JP. Scalable privacy-preserving distributed learning. Proc Priv Enhancing Technol. 2021;2021(2):323–47. doi: 10.2478/popets-2021-0030.
  8. Shi Z, Yang Z, Hassan A, Li F, Ding X. A privacy preserving federated learning scheme using homomorphic encryption and secret sharing. Telecommun Syst. 2023 Mar; 82(3): 419–33.
  9. Lansari M, Bellafqira R, Kapusta K, Thouvenot V, Bettan O, Coatrieux G. When federated learning meets watermarking: A comprehensive overview of techniques for intellectual property protection. Mach Learn Knowl Extr. 2023 Oct 4; 5(4): 1382–406.
  10. Brakerski Z, Vaikuntanathan V. Fully homomorphic encryption from ring-LWE and security for key dependent messages. In Annual cryptology conference. Berlin, Heidelberg: Springer Berlin Heidelberg; 2011 Aug 14; 505–524.
Regular Issue Subscription Original Research
Volume 03
Issue 02
Received 31/03/2025
Accepted 19/05/2025
Published 24/07/2025
Publication Time 115 Days
92

[first_name] [last_name]

My IP

PlumX Metrics