Exploration on the Development of Discipline Construction in Local University Libraries: Taking Shandong Technology and Business University as an Example

Abstract

Recent advancements have significantly improved automated task-solving capabilities using autonomous agents powered by large language models (LLMs). However, most LLMbased agents focus on dialogue, programming, or specialized domains, leaving gaps in addressing generative AI safety tasks. These gaps are primarily due to the challenges posed by LLM hallucinations and the lack of clear guidelines. In this paper, we propose Atlas, an advanced LLM-based multi-agent framework that integrates an efficient fuzzing workflow to target generative AI models, specifically focusing on jailbreak attacks against textto-image (T2I) models with safety filters. Atlas utilizes a visionlanguage model (VLM) to assess whether a prompt triggers the T2I model's safety filter. It then iteratively collaborates with both LLM and VLM to generate an alternative prompt that bypasses the filter. Atlas also enhances the reasoning abilities of LLMs in attack scenarios by leveraging multi-agent communication, in-context learning (ICL) memory mechanisms, and the chainof-thought (COT) approach. Our evaluation demonstrates that Atlas successfully jailbreaks several state-of-the-art T2I models in a black-box setting, which are equipped with multi-modal safety filters. In addition, Atlas outperforms existing methods in both query efficiency and the quality of the generated images.