Image de couverture de livre

Policy-Invisible Violations in LLM-Based Agents

Fuente: arXiv

Enregistré dans:

Détails bibliographiques
Auteurs principaux:	Wu, Jie, Gong, Ming
Format:	Preprint
Publié:	2026
Sujets:	Artificial Intelligence Computation and Language Cryptography and Security Machine Learning
Accès en ligne:	Acceder al recurso
Tags:	Ajouter un tag Pas de tags, Soyez le premier à ajouter un tag!

Documents similaires

Beyond Single-Agent Alignment: Preventing Context-Fragmented Violations in Multi-Agent Systems
par: Wu, Jie, et autres
Publié: (2026)

Federated In-Context LLM Agent Learning
par: Wu, Panlong, et autres
Publié: (2024)

BadAgent: Inserting and Activating Backdoor Attacks in LLM Agents
par: Wang, Yifei, et autres
Publié: (2024)

The Task Shield: Enforcing Task Alignment to Defend Against Indirect Prompt Injection in LLM Agents
par: Jia, Feiran, et autres
Publié: (2024)

Agent-ToM: Learning to Monitor Autonomous LLM Agents via Theory-of-Mind Reasoning
par: Ahmed, Nesreen K., et autres
Publié: (2026)

A Comprehensive Survey in LLM(-Agent) Full Stack Safety: Data, Training and Deployment
par: Wang, Kun, et autres
Publié: (2025)

STAC: When Innocent Tools Form Dangerous Chains to Jailbreak LLM Agents
par: Li, Jing-Jing, et autres
Publié: (2025)

Capability-Based Scaling Trends for LLM-Based Red-Teaming
par: Panfilov, Alexander, et autres
Publié: (2025)

Unlearned but Not Forgotten: Data Extraction after Exact Unlearning in LLM
par: Wu, Xiaoyu, et autres
Publié: (2025)

Measuring Real-World Prompt Injection Attacks in LLM-based Resume Screening
par: Zhang, Mohan, et autres
Publié: (2026)

Do Phone-Use Agents Respect Your Privacy?
par: Tang, Zhengyang, et autres
Publié: (2026)

Instructional Segment Embedding: Improving LLM Safety with Instruction Hierarchy
par: Wu, Tong, et autres
Publié: (2024)

IsolateGPT: An Execution Isolation Architecture for LLM-Based Agentic Systems
par: Wu, Yuhao, et autres
Publié: (2024)

TurboFuzzLLM: Turbocharging Mutation-based Fuzzing for Effectively Jailbreaking Large Language Models in Practice
par: Goel, Aman, et autres
Publié: (2025)

AdvPrefix: An Objective for Nuanced LLM Jailbreaks
par: Zhu, Sicheng, et autres
Publié: (2024)

Certifying LLM Safety against Adversarial Prompting
par: Kumar, Aounon, et autres
Publié: (2023)

Adaptive Instruction Composition for Automated LLM Red-Teaming
par: Zymet, Jesse, et autres
Publié: (2026)

Covert Malicious Finetuning: Challenges in Safeguarding LLM Adaptation
par: Halawi, Danny, et autres
Publié: (2024)

Efficient LLM Moderation with Multi-Layer Latent Prototypes
par: Chrabąszcz, Maciej, et autres
Publié: (2025)

LLM Ghostbusters: Surgical Hallucination Suppression via Adaptive Unlearning
par: Spracklen, Joseph, et autres
Publié: (2026)

REALISTA: Realistic Latent Adversarial Attacks that Elicit LLM Hallucinations
par: Liang, Buyun, et autres
Publié: (2026)

SECA: Semantically Equivalent and Coherent Attacks for Eliciting LLM Hallucinations
par: Liang, Buyun, et autres
Publié: (2025)

Systematically Analyzing Prompt Injection Vulnerabilities in Diverse LLM Architectures
par: Benjamin, Victoria, et autres
Publié: (2024)

LLM Cyber Evaluations Don't Capture Real-World Risk
par: Lukošiūtė, Kamilė, et autres
Publié: (2025)

Cross-Session Threats in AI Agents: Benchmark, Evaluation, and Algorithms
par: Azarafrooz, Ari
Publié: (2026)

Enhancing Prompt Injection Attacks to LLMs via Poisoning Alignment
par: Shao, Zedian, et autres
Publié: (2024)

From Domains to Instances: Dual-Granularity Data Synthesis for LLM Unlearning
par: Xu, Xiaoyu, et autres
Publié: (2026)

Are My Optimized Prompts Compromised? Exploring Vulnerabilities of LLM-based Optimizers
par: Zhao, Andrew, et autres
Publié: (2025)

SELF: A Robust Singular Value and Eigenvalue Approach for LLM Fingerprinting
par: Zhang, Hanxiu, et autres
Publié: (2025)

A Generative Approach to LLM Harmfulness Mitigation with Red Flag Tokens
par: Dobre, David, et autres
Publié: (2025)

Defending Against Alignment-Breaking Attacks via Robustly Aligned LLM
par: Cao, Bochuan, et autres
Publié: (2023)

BountyBench: Dollar Impact of AI Agent Attackers and Defenders on Real-World Cybersecurity Systems
par: Zhang, Andy K., et autres
Publié: (2025)

EIA: Environmental Injection Attack on Generalist Web Agents for Privacy Leakage
par: Liao, Zeyi, et autres
Publié: (2024)

Watch Out for Your Agents! Investigating Backdoor Threats to LLM-Based Agents
par: Yang, Wenkai, et autres
Publié: (2024)

There Are No Silly Questions: Evaluation of Offline LLM Capabilities from a Turkish Perspective
par: Yilmaz, Edibe, et autres
Publié: (2026)

In Vino Veritas and Vulnerabilities: Examining LLM Safety via Drunk Language Inducement
par: Shetty, Anudeex, et autres
Publié: (2026)

Tuning without Peeking: Provable Generalization Bounds and Robust LLM Post-Training
par: Labiad, Ismail, et autres
Publié: (2025)

Cheating Automatic LLM Benchmarks: Null Models Achieve High Win Rates
par: Zheng, Xiaosen, et autres
Publié: (2024)

RigorLLM: Resilient Guardrails for Large Language Models against Undesired Content
par: Yuan, Zhuowen, et autres
Publié: (2024)

AgentArmor: Enforcing Program Analysis on Agent Runtime Trace to Defend Against Prompt Injection
par: Wang, Peiran, et autres
Publié: (2025)