分析针对智能体 AI 系统的模型引导自动化攻击的防御性误导

Computer Science > Cryptography and Security arXiv:2606.20470 (cs) [Submitted on 18 Jun 2026 (v1), last revised 24 Sep 2026 (this version, v3)] Title:Analyzing Defensive Misdirection Against Model-Guided Automated Attacks on Agentic AI Systems Authors:Reza Soosahabi, Vivek Namsani View a PDF of the paper titled Analyzing Defensive Misdirection Against Model-Guided Automated Attacks on Agentic AI Systems, by Reza Soosahabi and 1 other authors View PDF HTML (experimental) Abstract:Agentic AI systems increasingly rely on language-model components to interpret instructions, process external data, invoke tools, and coordinate with other agents. These capabilities make prompt-injection and jailbreak attacks more consequential, especially as attackers adopt model-guided automation to scale probing, prompt refinement, and response evaluation. This work analyzes the resulting attack-defense setting through a probabilistic model of a target system, its defense mechanism, and the attacker's automated judge. Our analysis shows that conventional detect-and-block defenses can allow attacker success rate (ASR) to approach one as the query budget grows, since predictable refusals provide useful feedback to automated search. We then examine detect-and-misdirect, where detected malicious interactions receive controlled, non-operational responses designed to induce false-positive errors in the attacker's judge. This strategy reduces the positive predictive value of attacker-selected candidates and yields a bounded asymptotic ASR. We evaluate a proof-of-concept realization of this strategy through Contextual Misdirection via Progressive Engagement (CMPE), a lightweight conversational misdirection method designed to replace predictable refusal text with safe but strategically misleading responses in automated jailbreak settings. On jailbreak benchmarks, CMPE reduces estimated ASR upper bounds by up to two orders of magnitude and nearly eliminates verified attack success in end-to-end experiments with PAIR, GPTFuzz, and AutoDAN-Turbo. Comments: Subjects: Cryptography and Security (cs.CR); Artificial Intelligence (cs.AI) Cite as: arXiv:2606.20470 [cs.CR] (or arXiv:2606.20470v3 [cs.CR] for this version) https://doi.org/10.48550/arXiv.2606.20470 Focus to learn more arXiv-issued DOI via DataCite Submission history From: Reza Soosahabi [view email] [v1] Thu, 18 Jun 2026 16:50:28 UTC (974 KB) [v2] Fri, 26 Jun 2026 21:12:34 UTC (988 KB) [v3] Thu, 24 Sep 2026 14:46:04 UTC (1,063 KB) Full-text links: Access Paper: View a PDF of the paper titled Analyzing Defensive Misdirection Against Model-Guided Automated Attacks on Agentic AI Systems, by Reza Soosahabi and 1 other authors View PDF HTML (experimental) TeX Source view license Current browse context: cs.CR new | recent | 2026-06 Change to browse by: cs cs.AI References & Citations NASA ADS Google Scholar Semantic Scholar export BibTeX citation Loading... BibTeX formatted citation × loading... Data provided by: 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 Code, Data and Media Associated with this Article alphaXiv Toggle alphaXiv (What is alphaXiv?) Links to Code Toggle CatalyzeX Code Finder for Papers (What is CatalyzeX?) DagsHub Toggle DagsHub (What is DagsHub?) GotitPub Toggle Gotit.pub (What is GotitPub?) Huggingface Toggle Hugging Face (What is Huggingface?) ScienceCast Toggle ScienceCast (What is ScienceCast?) Demos Demos Replicate Toggle Replicate (What is Replicate?) Spaces Toggle Hugging Face Spaces (What is Spaces?) Spaces Toggle TXYZ.AI (What is TXYZ.AI?) Related Papers Recommenders and Search Tools Link to Influence Flower Influence Flower (What are Influence Flowers?) Core recommender toggle CORE Recommender (What is CORE?) Author Venue Institution Topic About arXivLabs arXivLabs: experimental projects with community collaborators arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website. Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them. Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs. Which authors of this paper are endorsers? | Disable MathJax (What is MathJax?)