#Adversarial Attacks

9개의 포스트

[논문리뷰] MUSE: A Run-Centric Platform for Multimodal Unified Safety Evaluation of Large Language Models

Yiran Chen이 arXiv에 게시한 'MUSE: A Run-Centric Platform for Multimodal Unified Safety Evaluation of Large Language Models' 논문에 대한 자세한 리뷰입니다.

#Review #Multimodal LLMs #Safety Evaluation #Red Teaming #Adversarial Attacks #Modality Switching #LLM Alignment #Compliance #ASR

2026년 3월 4일

[논문리뷰] Visual Memory Injection Attacks for Multi-Turn Conversations

Matthias Hein이 arXiv에 게시한 'Visual Memory Injection Attacks for Multi-Turn Conversations' 논문에 대한 자세한 리뷰입니다.

#Review #LVLM #Adversarial Attacks #Multi-Turn Conversations #Visual Memory Injection #Stealthy Attacks #Benign Anchoring #Context-Cycling

2026년 2월 18일

[논문리뷰] Few Tokens Matter: Entropy Guided Attacks on Vision-Language Models

arXiv에 게시된 'Few Tokens Matter: Entropy Guided Attacks on Vision-Language Models' 논문에 대한 자세한 리뷰입니다.

#Review #Vision-Language Models #Adversarial Attacks #Entropy-Guided Attacks #Token Vulnerability #Harmful Content #Cross-Model Transferability #Autoregressive Generation

2026년 1월 8일

[논문리뷰] M-ErasureBench: A Comprehensive Multimodal Evaluation Benchmark for Concept Erasure in Diffusion Models

Jun-Cheng Chen이 arXiv에 게시한 'M-ErasureBench: A Comprehensive Multimodal Evaluation Benchmark for Concept Erasure in Diffusion Models' 논문에 대한 자세한 리뷰입니다.

#Review #Diffusion Models #Concept Erasure #Multimodal Evaluation #Adversarial Attacks #Robustness #Textual Inversion #Latent Inversion #Cross-Attention

2026년 1월 5일

[논문리뷰] Pay Less Attention to Function Words for Free Robustness of Vision-Language Models

arXiv에 게시된 'Pay Less Attention to Function Words for Free Robustness of Vision-Language Models' 논문에 대한 자세한 리뷰입니다.

#Review #Vision-Language Models #Adversarial Robustness #Function Words #Cross-Attention #Adversarial Attacks #Differential Attention #Vision-Language Alignment

2025년 12월 10일

[논문리뷰] Hail to the Thief: Exploring Attacks and Defenses in Decentralised GRPO

arXiv에 게시된 'Hail to the Thief: Exploring Attacks and Defenses in Decentralised GRPO' 논문에 대한 자세한 리뷰입니다.

#Review #Decentralized RL #GRPO #LLM Post-training #Adversarial Attacks #Data Poisoning #Defense Mechanisms #In-context Attack #Out-of-context Attack

2025년 11월 13일

[논문리뷰] The Alignment Waltz: Jointly Training Agents to Collaborate for Safety

arXiv에 게시된 'The Alignment Waltz: Jointly Training Agents to Collaborate for Safety' 논문에 대한 자세한 리뷰입니다.

#Review #LLM Safety #Multi-agent Reinforcement Learning #Safety Alignment #Overrefusal #Adversarial Attacks #Feedback Agent #Conversation Agent #Dynamic Improvement Reward

2025년 10월 10일

[논문리뷰] WAInjectBench: Benchmarking Prompt Injection Detections for Web Agents

Neil Zhenqiang Gong이 arXiv에 게시한 'WAInjectBench: Benchmarking Prompt Injection Detections for Web Agents' 논문에 대한 자세한 리뷰입니다.

#Review #Prompt Injection #Web Agents #Multimodal AI #Adversarial Attacks #Detection Benchmarking #Large Language Models #Image-based Detection #Text-based Detection

2025년 10월 6일

[논문리뷰] Jailbreaking Commercial Black-Box LLMs with Explicitly Harmful Prompts

Liming Fang이 arXiv에 게시한 'Jailbreaking Commercial Black-Box LLMs with Explicitly Harmful Prompts' 논문에 대한 자세한 리뷰입니다.

#Review #LLM Jailbreaking #Red Teaming #Malicious Content Detection #Developer Messages #D-Attack #DH-CoT #Adversarial Attacks #Dataset Cleaning

2025년 8월 25일