LLM-Powered Automated Attacks

Girászi, Tamás and Papp, Natália and Oláh, Norbert and Huszti, Andrea (2026) LLM-Powered Automated Attacks. In: Proceedings of the 13th International Conference on Applied Informatics. Líceum Kiadó, Eger, pp. 107-121. ISBN 9789634963271

Preview

Text ICAI2026-pp107-121.pdf - Published Version Download (774kB) | Preview

Abstract

The increasing integration of large language models (LLMs) into systems introduces new attack surfaces that extend beyond traditional software vulnerabilities. While LLMs are commonly protected by prompt-level security mechanisms, recent researches show that these controls can be bypassed through carefully crafted inputs. In this paper, we propose an interaction model based on a Generative Adversarial Network (GAN) conceptual analogy and a dual-LLM framework for systematically examining and testing the security boundaries of LLMs through malicious code generation. The framework consists of two LLMs that iteratively produce attack-oriented prompts, interpret and implement them. Experimental results demonstrate that the proposed approach can generate outputs that are similar to realworld attack patterns, such as SQL injection and cross-site scripting. This research highlights the importance of LLM security controls and emphasizes the need for proactive, automated evaluation methods to improve the robustness and governance of LLM-based systems.

Actions (login required)

View Item