Copertina

Bypassing LLM Guardrails: An Empirical Analysis of Evasion Attacks against Prompt Injection and Jailbreak Detection Systems

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Hackett, William, Birch, Lewis, Trawicki, Stefan, Suri, Neeraj, Garraghan, Peter
Natura:	Preprint
Pubblicazione:	2025
Soggetti:	Cryptography and Security Artificial Intelligence Machine Learning I.2.7
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

Evaluating the Robustness of Large Language Model Safety Guardrails Against Adversarial Attacks
di: Young, Richard J.
Pubblicazione: (2025)

Detecting Prompt Injection Attacks Against Application Using Classifiers
di: Shaheer, Safwan, et al.
Pubblicazione: (2025)

Blind Spots in the Guard: How Domain-Camouflaged Injection Attacks Evade Detection in Multi-Agent LLM Systems
di: Pai, Aaditya
Pubblicazione: (2026)

Beyond the Benchmark: Innovative Defenses Against Prompt Injection Attacks
di: Shaheer, Safwan, et al.
Pubblicazione: (2025)

Jailbreaking Attacks vs. Content Safety Filters: How Far Are We in the LLM Safety Arms Race?
di: Xin, Yuan, et al.
Pubblicazione: (2025)

Super Suffixes: Bypassing Text Generation Alignment and Guard Models Simultaneously
di: Adiletta, Andrew, et al.
Pubblicazione: (2025)

PromptSAM+: Malware Detection based on Prompt Segment Anything Model
di: Wei, Xingyuan, et al.
Pubblicazione: (2024)

Defending against Backdoor Attacks via Module Switching
di: Li, Weijun, et al.
Pubblicazione: (2025)

Mitigating Trojanized Prompt Chains in Educational LLM Use Cases: Experimental Findings and Detection Tool Design
di: Charles, Richard M., et al.
Pubblicazione: (2025)

Temporal Attack Pattern Detection in Multi-Agent AI Workflows: An Open Framework for Training Trace-Based Security Models
di: Del Rosario, Ron F.
Pubblicazione: (2025)

DWFS-Obfuscation: Dynamic Weighted Feature Selection for Robust Malware Familial Classification under Obfuscation
di: Wei, Xingyuan, et al.
Pubblicazione: (2025)

POISONCRAFT: Practical Poisoning of Retrieval-Augmented Generation for Large Language Models
di: Shao, Yangguang, et al.
Pubblicazione: (2025)

How Few-shot Demonstrations Affect Prompt-based Defenses Against LLM Jailbreak Attacks
di: Wang, Yanshu, et al.
Pubblicazione: (2026)

ConfusionPrompt: Practical Private Inference for Online Large Language Models
di: Mai, Peihua, et al.
Pubblicazione: (2023)

UniC-RAG: Universal Knowledge Corruption Attacks to Retrieval-Augmented Generation
di: Geng, Runpeng, et al.
Pubblicazione: (2025)

Real AI Agents with Fake Memories: Fatal Context Manipulation Attacks on Web3 Agents
di: Patlan, Atharv Singh, et al.
Pubblicazione: (2025)

Prompted Contextual Vectors for Spear-Phishing Detection
di: Nahmias, Daniel, et al.
Pubblicazione: (2024)

sudoLLM: On Multi-role Alignment of Language Models
di: Saha, Soumadeep, et al.
Pubblicazione: (2025)

Efficient LLM Safety Evaluation through Multi-Agent Debate
di: Lin, Dachuan, et al.
Pubblicazione: (2025)

Accelerating Suffix Jailbreak attacks with Prefix-Shared KV-cache
di: Wang, Xinhai, et al.
Pubblicazione: (2026)

Reducing Information Overload: Because Even Security Experts Need to Blink
di: Kuehn, Philipp, et al.
Pubblicazione: (2022)

GuardVal: Dynamic Large Language Model Jailbreak Evaluation for Comprehensive Safety Testing
di: Zhang, Peiyan, et al.
Pubblicazione: (2025)

Train to Defend: First Defense Against Cryptanalytic Neural Network Parameter Extraction Attacks
di: Kurian, Ashley, et al.
Pubblicazione: (2025)

Can Watermarked LLMs be Identified by Users via Crafted Prompts?
di: Liu, Aiwei, et al.
Pubblicazione: (2024)

MarkLLM: An Open-Source Toolkit for LLM Watermarking
di: Pan, Leyi, et al.
Pubblicazione: (2024)

AgentSentry: Mitigating Indirect Prompt Injection in LLM Agents via Temporal Causal Diagnostics and Context Purification
di: Zhang, Tian, et al.
Pubblicazione: (2026)

Kill-Chain Canaries: Stage-Level Tracking of Prompt Injection Across Attack Surfaces and Model Safety Tiers
di: Wang, Haochuan Kevin, et al.
Pubblicazione: (2026)

PoTS: Proof-of-Training-Steps for Backdoor Detection in Large Language Models
di: Seddik, Issam, et al.
Pubblicazione: (2025)

SecEmb: Sparsity-Aware Secure Federated Learning of On-Device Recommender System with Large Embedding
di: Mai, Peihua, et al.
Pubblicazione: (2025)

Split-and-Denoise: Protect large language model inference with local differential privacy
di: Mai, Peihua, et al.
Pubblicazione: (2023)

Operationalizing a Threat Model for Red-Teaming Large Language Models (LLMs)
di: Verma, Apurv, et al.
Pubblicazione: (2024)

AI Safeguards, Generative AI and the Pandora Box: AI Safety Measures to Protect Businesses and Personal Reputation
di: Kumar, Prasanna
Pubblicazione: (2026)

Safeguarding Efficacy in Large Language Models: Evaluating Resistance to Human-Written and Algorithmic Adversarial Prompts
di: Downey-Webb, Tiarnaigh, et al.
Pubblicazione: (2025)

Predicting Known Vulnerabilities from Attack Descriptions Using Sentence Transformers
di: Othman, Refat
Pubblicazione: (2026)

Prompt Fencing: A Cryptographic Approach to Establishing Security Boundaries in Large Language Model Prompts
di: Peh, Steven
Pubblicazione: (2025)

JavelinGuard: Low-Cost Transformer Architectures for LLM Security
di: Datta, Yash, et al.
Pubblicazione: (2025)

VeriGuard: Enhancing LLM Agent Safety via Verified Code Generation
di: Miculicich, Lesly, et al.
Pubblicazione: (2025)

Lightweight LLMs for Network Attack Detection in IoT Networks
di: Sudasinghe, Piyumi Bhagya, et al.
Pubblicazione: (2026)

A Validated Prompt Bank for Malicious Code Generation: Separating Executable Weapons from Security Knowledge in 1,554 Consensus-Labeled Prompts
di: Young, Richard J., et al.
Pubblicazione: (2026)

The Quantum State Continuity Problem and Temporal Enforcement Against Fork Attacks
di: Ünsal, Samet
Pubblicazione: (2025)