mlsanityData to 7 Oct 2026

Seeing is Deceiving: Exploitation of Visual Pathways in Multi-Modal Language Models

arXiv 2411.05056 ↗First posted 7 Nov 2024cs.CR cs.AIcited by 4

On the map: Multimodal LLMs / vision-language models → MLLM jailbreaks and adversarial robustness

Abstract (arXiv)

Multi-Modal Language Models (MLLMs) have transformed artificial intelligence by combining visual and text data, making applications like image captioning, visual question answering, and multi-modal content creation possible. This ability to understand and work with complex information has made MLLMs useful in areas such as healthcare, autonomous systems, and digital content. However, integrating multiple types of data also creates security risks. Attackers can manipulate either the visual or text inputs, or both, to make the model produce unintended or even harmful responses. This paper reviews how visual inputs in MLLMs can be exploited by various attack strategies. We break down these attacks into categories: simple visual tweaks and cross-modal manipulations, as well as advanced strategies like VLATTACK, HADES, and Collaborative Multimodal Adversarial Attack (Co-Attack). These attacks can mislead even the most robust models while looking nearly identical to the original visuals, making them hard to detect. We also discuss the broader security risks, including threats to privacy and safety in important applications. To counter these risks, we review current defense methods like the SmoothVLM framework, pixel-wise randomization, and MirrorCheck, looking at their strengths and limitations. We also discuss new methods to make MLLMs more secure, including adaptive defenses, better evaluation tools, and security approaches that protect both visual and text data. By bringing together recent developments and identifying key areas for improvement, this review aims to support the creation of more secure and reliable multi-modal AI systems for real-world use.

Key facts. Cited by 4 papers in the map (data through 7 Oct 2026). Placed in MLLM jailbreaks and adversarial robustness, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 49 of its 61 references that are in the map. First posted on arXiv 7 Nov 2024.

Builds on

49 of 61 references.

Closest ancestors · 20

Same direction, newest first

  1. 2024Efficient and Effective Universal Adversarial Attack against Vision-Language Pre-training Models15 Oct
  2. 2024Break the Visual Perception: Adversarial Attacks Targeting Encoded Visual Tokens of Large Vision-Language Models9 Oct
  3. 2024AnyAttack: Towards Large-scale Self-supervised Adversarial Attacks on Vision-language Models7 Oct
  4. 2024Probing the Robustness of Vision-Language Pretrained Models: A Multimodal Adversarial Attack Approach24 Aug
  5. 2024Cross-modality Information Check for Detecting Jailbreaking in Multimodal Large Language Models31 Jul
  6. 2024Sim-CLIP: Unsupervised Siamese Adversarial Fine-Tuning for Robust and Semantically-Rich Vision-Language Models20 Jul
  7. 2024MirrorCheck: Efficient Adversarial Defense for Vision-Language Models13 Jun
  8. 2024Safeguarding Vision-Language Models Against Patched Visual Prompt Injectors17 May
  9. 2024Revisiting the Adversarial Robustness of Vision Language Models: a Multimodal Perspective30 Apr
  10. 2024Unbridled Icarus: A Survey of the Potential Perils of Image Inputs in Multimodal Large Language Model Security8 Apr
  11. 2024Images are Achilles' Heel of Alignment: Exploiting Visual Vulnerabilities for Jailbreaking Multimodal Large Language Models14 Mar
  12. 2024AdaShield: Safeguarding Multimodal Large Language Models from Structure-based Attack via Adaptive Shield Prompting14 Mar
8 more
  1. 2024Safety of Multimodal Large Language Models on Images and Texts1 Feb
  2. 2024MLLM-Protector: Ensuring MLLM's Safety without Hurting Performance5 Jan
  3. 2023OT-Attack: Enhancing Adversarial Transferability of Vision-Language Models via Optimal Transport Optimization7 Dec
  4. 2023On the Robustness of Large Multimodal Models Against Image Adversarial Attacks6 Dec
  5. 2023InstructTA: Instruction-Tuned Targeted Attack for Large Vision-Language Models4 Dec
  6. 2023MM-SafetyBench: A Benchmark for Safety Evaluation of Multimodal Large Language Models29 Nov
  7. 2023Visual Adversarial Examples Jailbreak Aligned Large Language Models22 Jun
  8. 2022Towards Adversarial Attack on Vision-Language Pre-training Models19 Jun

Other references · 29

  1. 2023Visual Instruction Tuningcited by 7,792
  2. 2014Explaining and Harnessing Adversarial Examplescited by 7,287
  3. 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 2,914
  4. 2023Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyondcited by 2,179
  5. 2020Extracting Training Data from Large Language Modelscited by 2,144
  6. 2023MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Modelscited by 1,624
  7. 2021Perceiver: General Perception with Iterative Attentioncited by 1,089
  8. 2023mPLUG-Owl: Modularization Empowers Large Language Models with Multimodalitycited by 1,038
  9. 2024What matters when building vision-language models?cited by 321
  10. 2023Diffusion Models for Imperceptible and Transferable Adversarial Attackcited by 93
  11. 2023VLATTACK: Multimodal Adversarial Attacks on Vision-Language Tasks via Pre-trained Modelscited by 66
  12. 2024VHELM: A Holistic Evaluation of Vision Language Modelscited by 53
17 more
  1. 2024The Curse of Multi-Modalities: Evaluating Hallucinations of Large Multimodal Models across Language, Visual, and Audiocited by 48
  2. 2024Cross-Modality Perturbation Synergy Attack for Person Re-identificationcited by 45
  3. 2024Jailbreaking and Mitigation of Vulnerabilities in Large Language Modelscited by 33
  4. 2023Adversarial Illusions in Multi-Modal Embeddingscited by 28
  5. 2023CosPGD: an efficient white-box adversarial attack for pixel-wise prediction taskscited by 27
  6. 2024A Survey on Multimodal Benchmarks: In the Era of Large AI Modelscited by 27
  7. 2024Securing Large Language Models: Addressing Bias, Misinformation, and Prompt Attackscited by 24
  8. 2024VisionGPT: Vision-Language Understanding Agent Using Generalized Multimodal Frameworkcited by 20
  9. 2024Surveying the MLLM Landscape: A Meta-Review of Current Surveyscited by 20
  10. 2021Sparse Adversarial Video Attacks with Spatial Transformationscited by 11
  11. 2020FAWA: Fast Adversarial Watermark Attack on Optical Character Recognition (OCR) Systemscited by 7
  12. 2024Systematic Categorization, Construction and Evaluation of New Attacks against Multi-modal Mobile GUI Agentscited by 5
  13. 2024Transcending Adversarial Perturbations: Manifold-Aided Adversarial Examples with Legitimate Semanticscited by 4
  14. 2024PG-Attack: A Precision-Guided Adversarial Attack Framework Against Vision Foundation Models for Autonomous Drivingcited by 4
  15. 2022Compound Tokens: Channel Fusion for Vision-Language Representation Learningcited by 2
  16. 2023Semantically Grounded QFormer for Efficient Vision Language Understandingcited by 2
  17. 2023Vulnerability Analysis of Transformer-based Optical Character Recognition to Adversarial Attackscited by 2

Built on by

Cited by year

Citing papers per year: 2025: 4.

4 papers.

  1. 2025Towards Trustworthy GUI Agents: A Surveycited by 33
  2. 2025Pushing the Limits of Safety: A Technical Report on the ATLAS Challenge 2025cited by 11
  3. 2025Hydra: An Agentic Reasoning Approach for Enhancing Adversarial Robustness and Mitigating Hallucinations in Vision-Language Modelscited by 2
  4. 2025ORCA: An Agentic Reasoning Framework for Hallucination and Adversarial Robustness in Vision-Language Modelscited by 0

Similar, not cited

Close in wording, no citation either way

  1. 2025Investigating Vulnerabilities and Defenses Against Audio-Visual Attacks: A Comprehensive Survey Emphasizing Multimodal Models
  2. 2026Adversarial Attacks on Multimodal Large Language Models: A Comprehensive Survey
  3. 2025When Data Manipulation Meets Attack Goals: An In-depth Survey of Attacks for VLMs
  4. 2026Leave My Images Alone: Preventing Multi-Modal Large Language Models from Analyzing Images via Visual Prompt Injection
  5. 2024A Survey of Attacks on Large Vision-Language Models: Resources, Advances, and Future Trends
  6. 2025Attention! Your Vision Language Model Could Be Maliciously Manipulated
  7. 2026Unveiling the Fragility of Vision-Language Models: Multi-Modal Adversarial Synergy via Texture-Constrained Perturbations and Cross-Modal Optimization
  8. 2026Covert Visual Prompt Injection against Commercial Multimodal Large Language Models