Disentangling Adversarial Prompts: A Semantic-Graph Defense for Robust LLM Security

Computer Science > Cryptography and Security [Submitted on 27 May 2026] Disentangling Adversarial Prompts: A Semantic-Graph Defense for Robust LLM Security Xiang Fang, Wanlong Fang Large Language Models (LLMs) are increasingly vulnerable to adversarial prompts that exploit semantic ambiguities to bypass safety mechanisms, resulting in harmful or inappropriate outputs. Such attacks, including jailbreaking and prompt injection, pose significant risks to the integrity and availability of LLMs in security-critical applications. This paper proposes the Adversarial Prompt Disentanglement (APD) framework, a novel defense mechanism that proactively identifies and neutralizes malicious components in input prompts before they are processed by the LLM. The APD framework integrates three key innovations: (1) a mutual information-based semantic decomposition method to isolate adversarial and benign prompt components, ensuring statistical independence; (2) a graph-based intent classification approach that leverages spectral analysis to detect malicious patterns in prompt semantics; and (3) a lightweight transformer-based classifier trained on real-world datasets of toxic and jailbreaking prompts, enabling efficient and accurate adversarial intent detection. Evaluated on diverse datasets containing adversarial prompts, APD demonstrates superior robustness, reducing harmful output generation by over 85\% while maintaining negligible impact on model performance. The framework's computational efficiency supports real-time deployment, making it a practical solution for securing LLMs. Our work addresses critical challenges in machine learning security on novel attacks and integrity methods for ML systems, and offers a scalable, ethically grounded defense against prompt-based adversarial threats. Comments: Published in AAAI 2026 Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI); Computer Vision and Pattern Recognition (cs.CV) Cite as: arXiv:2605.27823 [cs.CR]   (or arXiv:2605.27823v1 [cs.CR] for this version)   https://doi.org/10.48550/arXiv.2605.27823 Focus to learn more Submission history From: Xiang Fang [view email] [v1] Wed, 27 May 2026 01:30:06 UTC (1,406 KB) Access Paper: HTML (experimental) view license Current browse context: cs.CR < prev   |   next > new | recent | 2026-05 Change to browse by: cs cs.AI cs.CV References & Citations NASA ADS Google Scholar Semantic Scholar Export BibTeX Citation Bookmark Bibliographic Tools Bibliographic and Citation Tools Bibliographic Explorer Toggle Bibliographic Explorer (What is the Explorer?) Connected Papers Toggle Connected Papers (What is Connected Papers?) Litmaps Toggle Litmaps (What is Litmaps?) scite.ai Toggle scite Smart Citations (What are Smart Citations?) Code, Data, Media Demos Related Papers About arXivLabs Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)