Domain-camouflaged injection attacks embed malicious instructions in retrieved content using domain-appropriate vocabulary, evading standard detectors that rely on syntactic injection markers. When detection fails, practitioners need to know which defense architectures reduce attack success. We evaluate five prompting-based defenses (spotlighting, paraphrasing, prompt sandwiching, and two combinations) against domain-camouflaged injection across three model families (Claude Haiku, Llama 3.1 8B, Gemini 2.0 Flash) and three deployment domains (financial, legal, general) using 3,510 trials. Paraphrasing retrieved content before agent processing is the most consistently effective defense in this benchmark, reducing camouflage attack success rate by 55-84\% depending on model, and achieves lower attack success rates than our Llama Guard 4 configuration on every model tested. Defense effectiveness is strongly model-dependent: spotlighting halves attack success on Claude Haiku but provides no benefit on Llama 3.1 8B. Financial domain deployments face the highest residual risk at 26-33\% baseline attack success rate, with no prompting-based defense fully eliminating the threat on weaker models. These results provide the first systematic evaluation of prompting-based defenses specifically against camouflage-class injection attacks and establish benchmark-based recommendations for practitioners. All tasks use synthetically constructed professional documents; whether these benchmark rankings generalize to real enterprise documents remains an open question.
Mitigation of Camouflaged Adversarial Attacks in Autonomous Vehicles--A Case Study Using CARLA Simulator
Department of Computer Engineering Tennessee Technological University(计算机工程系田纳西技术大学);Department of Computer Science Tennessee Technological University(计算机科学系田纳西技术大学)
AdaShield: Safeguarding Multimodal Large Language Models from Structure-based Attack via Adaptive Shield Prompting
Peking University(北京大学);University of Wisconsin–Madison(威斯康星大学麦迪逊分校);International Digital Economy Academy(国际数字经济学院);University of California Davis(加州大学戴维斯分校)
Harmless Yet Harmful: Neutral Prompting Attacks for Stealthy Hallucination Steering in Agent Skills
Department of Computer Science(计算机科学系);National Yang Ming Chiao Tung University(阳明交通大学);Department of Electronics and Electrical Engineering(电子与电气工程系);School of Computing(计算学院);Institute of Science Tokyo(东京科学研究所)
CCFC: Core & Core-Full-Core Dual-Track Defense for LLM Jailbreak Protection
Department. of Math & Statistics, Boston University(数学与统计学系,波士顿大学);Department. of Computer Science, University at Albany(计算机科学系,阿尔巴尼大学);Department. of ECE & Systems Eng., Department. of Biomedical Eng., Faculty of Computing & Data Sciences, Boston University(电子工程与系统工程系,生物医学工程系,计算与数据科学学院,波士顿大学)
UC Berkeley(加州大学伯克利分校);UC Santa Barbara(加州大学圣巴巴拉分校);Duke University(杜克大学);National University of Singapore(新加坡国立大学);King Abdulaziz City for Science and Technology(国王阿卜杜勒阿齐兹城市科学技术学院);University of Washington(华盛顿大学)
Jailbreaking and Mitigation of Vulnerabilities in Large Language Models
Hunan University Changsha, PRC;Georgia Institute of Technology Atlanta, USA;Kyoto University Kyoto, Japan;Purdue University West Lafayette, USA;National Taiwan Normal University Taipei, ROC;University of Liverpool Suzhou, PRC;Hong Kong University of Science;University of Hawaii Honolulu, USA;The University of Texas at Dallas Dallas, USA;University of Wisconsin-Madison Madison, USA;Emory University Atlanta, USA;College of William \& Mary Williamsburg, USA