Semantic-Preserving Adversarial Attacks on LLMs: An Adaptive Greedy Binary Search Approach

ACL ARR 2025 May Submission187 Authors

Abstract: Large Language Models (LLMs) increasingly rely on automatic prompt engineering in graphical user interfaces (GUIs) to refine user inputs and enhance response accuracy. However, the diversity of user requirements often leads to unintended misinterpretations, where automated optimizations distort original intentions and produce erroneous outputs. To address this challenge, we propose the Adaptive Greedy Binary Search (AGBS) method, which simulates common prompt optimization mechanisms while preserving semantic stability. Our approach dynamically evaluates the impact of such strategies on LLM performance, enabling robust adversarial sample generation. Through extensive experiments on open and closed-source LLMs, we demonstrate AGBS’s effectiveness in balancing semantic consistency and attack efficacy. Our findings offer actionable insights for designing more reliable prompt optimization systems. Code is available at: https://anonymous.4open.science/r/A5E7202F .

Paper Type: Long

Research Area: Ethics, Bias, and Fairness

Research Area Keywords: Large Language Models, Automatic Prompt Engineering, Adversarial Attack

Contribution Types: Model analysis & interpretability, NLP engineering experiment

Languages Studied: English

Submission Number: 187