Modal Attack

Modal attacks exploit vulnerabilities in machine learning models that process multiple data modalities (e.g., image and text, or LiDAR and electromagnetic signals). Current research focuses on developing and evaluating these attacks across various model architectures, including vision-language models and those used in autonomous driving, often leveraging techniques like adversarial perturbations and backdoor injections to manipulate model outputs. Understanding and mitigating these attacks is crucial for ensuring the safety and reliability of increasingly prevalent multimodal AI systems in diverse applications, from robotics to person re-identification. The field is actively exploring both the effectiveness of different attack strategies and the development of robust defenses.

Papers

November 3, 2024

UniGuard: Towards Universal Safety Guardrails for Jailbreak Attacks on Multimodal Large Language Models
Sejoon Oh, Yiqiao Jin, Megha Sharma, Donghyun Kim, Eric Ma, Gaurav Verma, Srijan Kumar
Multimodal Large Language Model Jailbreak Attack Stronger Generalizability Guardrail Model Modal Attack

October 20, 2024

Jailbreaking and Mitigation of Vulnerabilities in Large Language Models
Benji Peng, Ziqian Bi, Qian Niu, Ming Liu, Pohsun Feng, Tianyang Wang, Lawrence K.Q. Yan, Yizhu Wen, Yichao Zhang, Caitlyn Heqi Yin
Multi Agent Jailbreak Attack Security Vulnerability Faulty Negative Mitigation Alignment Approach Modal Attack Adversarial Prompting

October 9, 2024

ETA: Evaluating Then Aligning Safety of Vision Language Models at Inference Time
Yi Ding, Bolian Li, Ruqi Zhang
Vision Language Model Adversarial Input Inference Time Multimodal AI Temporal Alignment Safety Alignment Modal Attack Estimated Time of Arrival

October 3, 2024

BadCM: Invisible Backdoor Attack Against Cross-Modal Learning
Zheng Zhang, Xu Yuan, Lei Zhu, Jingkuan Song, Liqiang Nie
Cross Modal Modal Attack Man in the Middle Attack Invisible Backdoor Attack Multimodal Backdoor

September 26, 2024

PhantomLiDAR: Cross-modality Signal Injection Attacks against LiDAR
Zizhi Jin, Qinhong Jiang, Xuancun Lu, Chen Yan, Xiaoyu Ji, Wenyuan Xu
Mid Range LiDAR LiDAR Data LiDAR Based Modal Attack Phantom Threat Electromagnetic Signal Injection Attack

September 20, 2024

Manipulation Facing Threats: Evaluating Physical Vulnerabilities in End-to-End Vision Language Action Models
Hao Cheng, Erjia Xiao, Chengyuan Yu, Zhao Yao, Jiahang Cao, Qiang Zhang, Jiaxu Wang, Mengshu Sun, Kaidi Xu, Jindong Gu, Renjing Xu
Manipulation Task Strategic Manipulation Threat Word Vision Language Action Modal Attack

July 31, 2024

Vera Verto: Multimodal Hijacking Attack
Minxing Zhang, Ahmed Salem, Michael Backes, Yang Zhang
Modal Attack Multimodal Attack Hijacking Task Model Hijacking Attack

January 18, 2024

Cross-Modality Perturbation Synergy Attack for Person Re-identification
Yunpeng Gong, Zhun Zhong, Yansong Qu, Zhiming Luo, Rongrong Ji, Min Jiang
Person Re Identification Visual Modality Modal Attack Multimodal Re Identification

December 31, 2023

Balanced Multi-modal Federated Learning via Cross-Modal Infiltration
Yunfeng Fan, Wenchao Xu, Haozhao Wang, Jiaqi Zhu, Song Guo
Federated Learning Multimodal Data Modal Attack Modality Imbalance Modality Heterogeneity

December 20, 2023

Mutual-modality Adversarial Attack with Semantic Perturbation
Jingwen Ye, Ruonan Yu, Songhua Liu, Xinchao Wang
Adversarial Attack Adversarial Sample Transferable Attack Modal Attack Man in the Middle Attack Semantic Perturbation

October 16, 2023

Survey of Vulnerabilities in Large Language Models Revealed by Adversarial Attacks
Erfan Shayegani, Md Abdullah Al Mamun, Yu Fu, Pedram Zaree, Yue Dong, Nael Abu-Ghazaleh
Timely Survey Adversarial Attack Security Vulnerability Text Attack Modal Attack

October 7, 2023

VLATTACK: Multimodal Adversarial Attacks on Vision-Language Tasks via Pre-trained Models
Ziyi Yin, Muchao Ye, Tianrong Zhang, Tianyu Du, Jinguo Zhu, Han Liu, Jinghui Chen, Ting Wang, Fenglong Ma
Adversarial Attack Adversarial Robustness Pre Trained Model Vision Language Image Text Pair Adversarial Sample Vision Language Task Modal Attack

July 27, 2023

Unified Adversarial Patch for Visible-Infrared Cross-modal Attacks in the Physical World
Xingxing Wei, Yao Huang, Yitong Sun, Jie Yu
Adversarial Attack Adversarial Patch Physical World Modal Attack

July 26, 2023

Jailbreak in pieces: Compositional Adversarial Attacks on Multi-Modal Language Models
Erfan Shayegani, Yue Dong, Nael Abu-Ghazaleh
Jailbreak Attack Adversarial Image Multi Modal Language Model Modal Attack Non Contiguous Piece Compositional Adversarial

July 15, 2023

Unified Adversarial Patch for Cross-modal Attacks in the Physical World
Xingxing Wei, Yao Huang, Yitong Sun, Jie Yu
Adversarial Attack Adversarial Patch Physical World Modal Attack

May 16, 2023

Iterative Adversarial Attack on Image-guided Story Ending Generation
Youze Wang, Wenbo Hu, Richang Hong
Adversarial Text Modal Attack Multimodal Attack Multimodal Text Generation

April 28, 2023

Fusion is Not Enough: Single Modal Attacks on Fusion Models for 3D Object Detection
Zhiyuan Cheng, Hongjun Choi, James Liang, Shiwei Feng, Guanhong Tao, Dongfang Liu, Michael Zuzak, Xiangyu Zhang
Adversarial Attack 3D Object Detection Hybrid Fusion Multi Sensor Fusion Modal Attack Fusion Model Camera LiDAR Fusion Model

October 3, 2022

Push-Pull: Characterizing the Adversarial Robustness for Audio-Visual Active Speaker Detection
Xuanjun Chen, Haibin Wu, Helen Meng, Hung-yi Lee, Jyh-Shing Roger Jang
Adversarial Robustness Active Speaker Detection Modal Attack Push Pull

June 17, 2022

Is Multi-Modal Necessarily Better? Robustness Evaluation of Multi-modal Fake News Detection
Jinyin Chen, Chengyu Jia, Haibin Zheng, Ruoxi Chen, Chenbo Fu
Fake News Detection Robustness Evaluation Modal Attack

December 10, 2021

Cross-Modal Transferable Adversarial Attacks from Images to Videos
Zhipeng Wei, Jingjing Chen, Zuxuan Wu, Yu-Gang Jiang
Adversarial Attack Adversarial Example Gameplay Video Transferable Adversarial Attack Adversarial Framework Modal Attack

Modal Attack

Papers

UniGuard: Towards Universal Safety Guardrails for Jailbreak Attacks on Multimodal Large Language Models

Jailbreaking and Mitigation of Vulnerabilities in Large Language Models

ETA: Evaluating Then Aligning Safety of Vision Language Models at Inference Time

BadCM: Invisible Backdoor Attack Against Cross-Modal Learning

PhantomLiDAR: Cross-modality Signal Injection Attacks against LiDAR

Manipulation Facing Threats: Evaluating Physical Vulnerabilities in End-to-End Vision Language Action Models

Vera Verto: Multimodal Hijacking Attack

Cross-Modality Perturbation Synergy Attack for Person Re-identification

Balanced Multi-modal Federated Learning via Cross-Modal Infiltration

Mutual-modality Adversarial Attack with Semantic Perturbation

Survey of Vulnerabilities in Large Language Models Revealed by Adversarial Attacks

VLATTACK: Multimodal Adversarial Attacks on Vision-Language Tasks via Pre-trained Models

Unified Adversarial Patch for Visible-Infrared Cross-modal Attacks in the Physical World

Jailbreak in pieces: Compositional Adversarial Attacks on Multi-Modal Language Models

Unified Adversarial Patch for Cross-modal Attacks in the Physical World

Iterative Adversarial Attack on Image-guided Story Ending Generation

Fusion is Not Enough: Single Modal Attacks on Fusion Models for 3D Object Detection

Push-Pull: Characterizing the Adversarial Robustness for Audio-Visual Active Speaker Detection

Is Multi-Modal Necessarily Better? Robustness Evaluation of Multi-modal Fake News Detection

Cross-Modal Transferable Adversarial Attacks from Images to Videos