Engineering Approaches to Improving Security and Reliability in Generative Artificial Intelligence Systems

Article Sidebar

PDF
Published: 26.08.2026

Main Article Content

Babayev Mahammad Murad
Baku Higher Oil School

Abstract

The rapid integration of generative artificial intelligence into software platforms, decision-support systems, information retrieval environments and autonomous applications has created a new class of engineering challenges. In addition to conventional software vulnerabilities, generative AI systems may produce factually incorrect information, follow maliciously injected instructions, reveal sensitive data, generate unsafe content or initiate unintended actions when connected to external tools. These characteristics make security and reliability inseparable requirements rather than independent design objectives. This article examines contemporary engineering approaches for improving the security and reliability of generative AI systems, with particular emphasis on large language model -based applications. The study adopts an engineering-oriented analytical review of recent research published mainly between 2024 and 2026. Prompt injection, jailbreaking, hallucination, excessive refusal, information leakage, unsafe tool use and component -level vulnerabilities are considered as major risk categories.

Downloads

Download data is not yet available.

Article Details

Data Availability Statement

All data supporting the findings of this study are presented in the text of the scientific work.

Issue

No. 307 (2026): 8th ISPC «Recent Advances in Global Science» (August 26-28, 2026; Vilnius, Lithuania).

Section

Information and Web technologies
Creative Commons License

This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.

Author Biography

Babayev Mahammad Murad, Baku Higher Oil School

Undergraduate Student in Computer Engineering

How to Cite

Babayev, M. (2026). Engineering Approaches to Improving Security and Reliability in Generative Artificial Intelligence Systems. Scientific Collection «InterConf», 307, 87–92. https://interconf.openpubarchive.com/index.php/proceeding/article/view/66

References

Bin, W., Jiazheng, Q., Yu, X., Hansen, H., Hao, Y., Gao, A., Wan, Z., Li, H., & Tsang, I. (2026). Safety Sidecar: Reflection-Driven Runtime Control for Safer Agents. Findings of the Association for Computational Linguistics: ACL 2026, pp. 30842–30856. Association for Computational Linguistics. https://doi.org/10.18653/v1/2026.findings-acl.1542

Chang, Z., Li, M., Huang, Y., Jiang, Z., Jia, X., Xiong, Q., Wang, J., Li, Z., & Wang, Q. (2026). Know Thy Enemy: Securing LLMs Against Prompt Injection via Diverse Data Synthesis and Instruction-Level Chain-of-Thought Learning. Findings of the Association for Computational Linguistics: ACL 2026, pp. 3540–3561. Association for Computational Linguistics. https://doi.org/10.18653/v1/2026.findings-acl.173

Wang, X., Jian, S., Li, S., Li, X., Li, Z., Ji, B., Wang, B., & Yu, J. (2026). JPU: Bridging Jailbreak Defense and Unlearning via On-Policy Path Rectification. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics, pp. 7661–7674. Association for Computational Linguistics. https://doi.org/10.18653/v1/2026.acl-long.348

Chen, Y., Li, H., Zheng, Z., Wu, D., Song, Y., & Hooi, B. (2025). Defense Against Prompt Injection Attack by Leveraging Attack Techniques. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics, pp. 18331–18347. Association for Computational Linguistics. https://doi.org/10.18653/v1/2025.acl- long.897

Li, H., Liu, X., Zhang, N., & Xiao, C. (2025). PIGuard: Prompt Injection Guardrail via Mitigating Overdefense for Free. Proceedings of the 63rd Annual Meeting of the Association for Computational Linguistics, pp. 30420–30437. Association for Computational Linguistics. https://doi.org/10.18653/v1/2025.acl-long.1468

Hackett, W., Birch, L., Trawicki, S., Suri, N., & Garraghan, P. (2025). Bypassing LLM Guardrails: An Empirical Analysis of Evasion Attacks against Prompt Injection and Jailbreak Detection Systems. Proceedings of the First Workshop on LLM Security (LLMSEC), pp. 101–114. Association for Computational Linguistics.

Wang, S., Wang, X., Mei, J., Xie, Y., Chen, S.-Q., & Xiong, W. (2025). Developing a Reliable, Fast, General-Purpose Hallucination Detection and Mitigation Service. Proceedings of NAACL 2025: Industry Track, pp. 971–978. Association for Computational Linguistics. https://doi.org/10.18653/v1/2025.naacl-industry.72

Xu, Z., Liu, Y., Deng, G., Li, Y., & Picek, S. (2024). A Comprehensive Study of Jailbreak Attack versus Defense for Large Language Models. Findings of the Association for Computational Linguistics: ACL 2024, pp. 7432–7449. Association for Computational Linguistics. https://doi.org/10.18653/v1/2024.findings-acl.443

Xie, Y., Fang, M., Pi, R., & Gong, N. (2024). GradSafe: Detecting Jailbreak Prompts for LLMs via Safety-Critical Gradient Analysis. Proceedings of the 62nd Annual Meeting of the Association for Computational Linguistics, pp. 507–518. Association for Computational Linguistics. https://doi.org/10.18653/v1/2024.acl-long.30

Zhao, W., Li, Z., Li, Y., Zhang, Y., & Sun, J. (2024). Defending Large Language Models Against Jailbreak Attacks via Layer-Specific Editing. Findings of the Association for Computational Linguistics: EMNLP 2024, pp. 5094–5109. Association for Computational Linguistics. https://doi.org/10.18653/v1/2024.findings-emnlp.293