Cover Image

Mirage in the Eyes: Hallucination Attack on Multi-modal Large Language Models with Only Attention Sink

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Wang, Yining, Zhang, Mi, Sun, Junjie, Wang, Chenyue, Yang, Min, Xue, Hui, Tao, Jialing, Duan, Ranjie, Liu, Jiexi
Format:	Preprint
Published:	2025
Subjects:	Machine Learning Cryptography and Security Computer Vision and Pattern Recognition
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

Mirage-in-the-Eyes: Hallucination Attack on Multi-modal Large Language Models with Only Attention Sink
by: Wang
Published: (2025)

A Causal Perspective for Enhancing Jailbreak Attack and Defense
by: Pan, Licheng, et al.
Published: (2026)

Jailbreaking Multimodal Large Language Models via Shuffle Inconsistency
by: Zhao, Shiji, et al.
Published: (2025)

Inducing Overthink: Hierarchical Genetic Algorithm-based DoS Attack on Black-Box Large Language Reasoning Models
by: Wang, Shuqiang, et al.
Published: (2026)

Mirage Fools the Ear, Mute Hides the Truth: Precise Targeted Adversarial Attacks on Polyphonic Sound Event Detection Systems
by: Su, Junjie, et al.
Published: (2025)

MRJ-Agent: An Effective Jailbreak Agent for Multi-Round Dialogue
by: Wang, Fengxiang, et al.
Published: (2024)

Learning to Detect Unseen Jailbreak Attacks in Large Vision-Language Models
by: Liang, Shuang, et al.
Published: (2025)

MirageBackdoor: A Stealthy Attack that Induces Think-Well-Answer-Wrong Reasoning
by: Zeng, Yizhe, et al.
Published: (2026)

Mirage: Defense against CrossPath Attacks in Software Defined Networks
by: Murtuza, Shariq, et al.
Published: (2024)

STARE: Step-wise Temporal Alignment and Red-teaming Engine for Multi-modal Toxicity Attack
by: Mao, Xutao, et al.
Published: (2026)

Heuristic-Induced Multimodal Risk Distribution Jailbreak Attack for Multimodal Large Language Models
by: Teng, Ma, et al.
Published: (2024)

PBI-Attack: Prior-Guided Bimodal Interactive Black-Box Jailbreak Attack for Toxicity Maximization
by: Cheng, Ruoxi, et al.
Published: (2024)

BadToken: Token-level Backdoor Attacks to Multi-modal Large Language Models
by: Yuan, Zenghui, et al.
Published: (2025)

Misrouter: Exploiting Routing Mechanisms for Input-Only Attacks on Mixture-of-Experts LLMs
by: Fei, Zekun, et al.
Published: (2026)

BELT: Old-School Backdoor Attacks can Evade the State-of-the-Art Defense with Backdoor Exclusivity Lifting
by: Qiu, Huming, et al.
Published: (2023)

Yet Another Mirage of Breaking MIRAGE: Debunking Occupancy-based Side-Channel Attacks on Fully Associative Randomized Caches
by: Cao, Chris, et al.
Published: (2025)

Dummy-Aware Weighted Attack (DAWA): Breaking the Safe Sink in Dummy Class Defenses
by: Yu, Yunrui, et al.
Published: (2026)

Preventing Non-intrusive Load Monitoring Privacy Invasion: A Precise Adversarial Attack Scheme for Networked Smart Meters
by: He, Jialing, et al.
Published: (2024)

ADCA: Attention-Driven Multi-Party Collusion Attack in Federated Self-Supervised Learning
by: Wang, Jiayao, et al.
Published: (2026)

HPE: Hallucinated Positive Entanglement for Backdoor Attacks in Federated Self-Supervised Learning
by: Wang, Jiayao, et al.
Published: (2026)

Ciphertext-Only Attack on a Secure $k$-NN Computation on Cloud
by: Murthy, Shyam, et al.
Published: (2024)

Systematic Categorization, Construction and Evaluation of New Attacks against Multi-modal Mobile GUI Agents
by: Yang, Yulong, et al.
Published: (2024)

HTS-Attack: Heuristic Token Search for Jailbreaking Text-to-Image Models
by: Gao, Sensen, et al.
Published: (2024)

Follow My Eyes: Backdoor Attacks on VLM-based Scanpath Prediction
by: Romero, Diana, et al.
Published: (2026)

AttackEval: A Systematic Empirical Study of Prompt Injection Attack Effectiveness Against Large Language Models
by: Wang, Jackson
Published: (2026)

Parasites in the Toolchain: A Large-Scale Analysis of Attacks on the MCP Ecosystem
by: Zhao, Shuli, et al.
Published: (2025)

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

Contextualizing Sink Knowledge for Java Vulnerability Discovery
by: Fleischer, Fabian, et al.
Published: (2026)

Learning-based Privacy-Preserving Graph Publishing Against Sensitive Link Inference Attacks
by: Wu, Yucheng, et al.
Published: (2025)

SEW: Strengthening Robustness of Black-box DNN Watermarking via Specificity Enhancement
by: Qiu, Huming, et al.
Published: (2026)

MMCert: Provable Defense against Adversarial Attacks to Multi-modal Models
by: Wang, Yanting, et al.
Published: (2024)

BM-PAW: A Profitable Mining Attack in the PoW-based Blockchain System
by: Hu, Junjie, et al.
Published: (2024)

Typographic Attacks in a Multi-Image Setting
by: Wang, Xiaomeng, et al.
Published: (2025)

Towards Label-Only Membership Inference Attack against Pre-trained Large Language Models
by: He, Yu, et al.
Published: (2025)

An Attack Method for Medical Insurance Claim Fraud Detection based on Generative Adversarial Network
by: Pang, Yining, et al.
Published: (2025)

Keeping an Eye on LLM Unlearning: The Hidden Risk and Remedy
by: Ren, Jie, et al.
Published: (2025)

Efficient Blockchain-based Steganography via Backcalculating Generative Adversarial Network
by: Chen, Zhuo, et al.
Published: (2025)

FlowSteer: Prompt-Only Workflow Steering Exposes Planning-Time Vulnerabilities in Multi-Agent LLM Systems
by: Li, Fanxiao, et al.
Published: (2026)

ICL-EVADER: Zero-Query Black-Box Evasion Attacks on In-Context Learning and Their Defenses
by: He, Ningyuan, et al.
Published: (2026)

EagleEye: Attention to Unveil Malicious Event Sequences from Provenance Graphs
by: Gysel, Philipp, et al.
Published: (2024)