Imagen de Portada

Servant, Stalker, Predator: How An Honest, Helpful, And Harmless (3H) Agent Unlocks Adversarial Skills

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autor principal:	Noever, David
Formato:	Preprint
Publicado:	2025
Materias:	Cryptography and Security Artificial Intelligence
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Language Models And A Second Opinion Use Case: The Pocket Professional
por: Noever, David
Publicado: (2024)

Hallucinating AI Hijacking Attack: Large Language Models and Malicious Code Recommenders
por: Noever, David, et al.
Publicado: (2024)

Towards Safe and Honest AI Agents with Neural Self-Other Overlap
por: Carauleanu, Marc, et al.
Publicado: (2024)

HarmfulSkillBench: How Do Harmful Skills Weaponize Your Agents?
por: Jiang, Yukun, et al.
Publicado: (2026)

Attention Masks Help Adversarial Attacks to Bypass Safety Detectors
por: Shi, Yunfan
Publicado: (2024)

Understanding LLM Behavior When Encountering User-Supplied Harmful Content in Harmless Tasks
por: Chu, Junjie, et al.
Publicado: (2026)

HonestCyberEval: An AI Cyber Risk Benchmark for Automated Software Exploitation
por: Ristea, Dan, et al.
Publicado: (2024)

SkillTester: Benchmarking Utility and Security of Agent Skills
por: Wang, Leye, et al.
Publicado: (2026)

BadSkill: Backdoor Attacks on Agent Skills via Model-in-Skill Poisoning
por: Tie, Guiyao, et al.
Publicado: (2026)

SkillJect: Effectively Automating Skill-Based Prompt Injection for Skill-Enabled Agents
por: Jia, Xiaojun, et al.
Publicado: (2026)

SkillTrojan: Backdoor Attacks on Skill-Based Agent Systems
por: Feng, Yunhao, et al.
Publicado: (2026)

Context Matters: Repository-Aware Security Analysis of the Agent Skill Ecosystem
por: Holzbauer, Florian, et al.
Publicado: (2026)

SkillSieve: A Hierarchical Triage Framework for Detecting Malicious AI Agent Skills
por: Hou, Yinghan, et al.
Publicado: (2026)

AgentTrap: Measuring Runtime Trust Failures in Third-Party Agent Skills
por: Zhuang, Haomin, et al.
Publicado: (2026)

Detecting Adversarial Fine-tuning with Auditing Agents
por: Egler, Sarah, et al.
Publicado: (2025)

Technical Report: Exploring the Emerging Threats of the Agent Skill Ecosystem
por: Beurer-Kellner, Luca, et al.
Publicado: (2026)

Structured Security Auditing and Robustness Enhancement for Untrusted Agent Skills
por: Lv, Lijia, et al.
Publicado: (2026)

No Attack Required: Semantic Fuzzing for Specification Violations in Agent Skills
por: Li, Ying, et al.
Publicado: (2026)

Living Off the LLM: How LLMs Will Change Adversary Tactics
por: Oesch, Sean, et al.
Publicado: (2025)

Towards Secure Agent Skills: Architecture, Threat Taxonomy, and Security Analysis
por: Li, Zhiyuan, et al.
Publicado: (2026)

Proteus: A Self-Evolving Red Team for Agent Skill Ecosystems
por: Zhou, Zhaojiacheng
Publicado: (2026)

RouteGuard: Internal-Signal Detection of Skill Poisoning in LLM Agents
por: Xiao, Wenjie, et al.
Publicado: (2026)

Latent Fusion Jailbreak: Blending Harmful and Harmless Representations to Elicit Unsafe LLM Outputs
por: Xing, Wenpeng, et al.
Publicado: (2025)

Adversarial Agents: Black-Box Evasion Attacks with Reinforcement Learning
por: Domico, Kyle, et al.
Publicado: (2025)

SkCC: Portable and Secure Skill Compilation for Cross-Framework LLM Agents
por: Ouyang, Yipeng, et al.
Publicado: (2026)

Credential Leakage in LLM Agent Skills: A Large-Scale Empirical Study
por: Chen, Zhihao, et al.
Publicado: (2026)

WARD: Adversarially Robust Defense of Web Agents Against Prompt Injections
por: Cao, Tri, et al.
Publicado: (2026)

How stealthy is stealthy? Studying the Efficacy of Black-Box Adversarial Attacks in the Real World
por: Panebianco, Francesco, et al.
Publicado: (2025)

ClawKeeper: Comprehensive Safety Protection for OpenClaw Agents Through Skills, Plugins, and Watchers
por: Liu, Songyang, et al.
Publicado: (2026)

Under the Hood of SKILL.md: Semantic Supply-chain Attacks on AI Agent Skill Registry
por: Saha, Shoumik, et al.
Publicado: (2026)

DECEPTICON: How Dark Patterns Manipulate Web Agents
por: Cuvin, Phil, et al.
Publicado: (2025)

Explanation as a Watermark: Towards Harmless and Multi-bit Model Ownership Verification via Watermarking Feature Attribution
por: Shao, Shuo, et al.
Publicado: (2024)

Doppelganger Method: Breaking Role Consistency in LLM Agent via Prompt-based Transferable Adversarial Attack
por: Kang, Daewon, et al.
Publicado: (2025)

When Safe Skills Collide: Measuring Compositional Risk in Agent Skill Ecosystems
por: Wang, Su, et al.
Publicado: (2026)

The Blind Spot of Agent Safety: How Benign User Instructions Expose Critical Vulnerabilities in Computer-Use Agents
por: Ding, Xuwei, et al.
Publicado: (2026)

Next-Generation Phishing: How LLM Agents Empower Cyber Attackers
por: Afane, Khalifa, et al.
Publicado: (2024)

Behavioral Integrity Verification for AI Agent Skills
por: Wu, Yuhao, et al.
Publicado: (2026)

AccLock: Unlocking Identity with Heartbeat Using In-Ear Accelerometers
por: Wang, Lei, et al.
Publicado: (2026)

From Allies to Adversaries: Manipulating LLM Tool-Calling through Adversarial Injection
por: Wang, Haowei, et al.
Publicado: (2024)

E-PhishGen: Unlocking Novel Research in Phishing Email Detection
por: Pajola, Luca, et al.
Publicado: (2025)