mlsanityData to 14 Oct 2026

Evolving Safety Landscape of Multi-modal Large Language Models: A Survey of Emerging Threats and Safeguards

arXiv 2608.07535 ↗First posted 27 Jul 2026cs.LG cs.AI cs.CYcited by 0

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

also in 1 more
Abstract (arXiv)

Multi-modal large language models (MLLMs) integrate heterogeneous modalities through modality alignment and fusion, enabling stronger understanding and reasoning. However, this architectural shift reshapes the safety landscape of machine learning. Increased model complexity and cross-modal interactions give rise to novel threats, including compromised modality integration, modality misalignment, and fused safety risks, reflecting shifts in threat modeling beyond uni-modal assumptions. These shifts, in turn, impose new constraints on safety solutions not captured by existing frameworks rooted in uni-modal learning. Motivated by these challenges, this survey provides a systematic analysis of the evolving safety landscape of MLLMs. We first propose a multimodal grounded taxonomy of safety threats and analyze shifts in threat models, covering adversarial attacks, data poisoning, jailbreaks, and hallucinations. We then summarize updated safety assumptions and organize recent advances in MLLM safety strategies accordingly. Finally, we discuss open challenges and future directions to inform the development of more principled and scalable safety mechanisms for multimodal systems.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in MLLM jailbreaks and adversarial robustness, one of 16 directions in Multimodal LLMs / vision-language models. Builds on 144 of its 174 references that are in the map. First posted on arXiv 27 Jul 2026.

Builds on

144 of 174 references.

Closest ancestors · 46

Same direction, newest first

  1. 2025Self-Aware Safety Augmentation: Leveraging Internal Semantic Understanding to Enhance Safety in Vision-Language Models29 Jul
  2. 2025SafePTR: Token-Level Jailbreak Defense in Multimodal LLMs via Prune-then-Restore Mechanism2 Jul
  3. 2025Backdoor Attack on Vision Language Models with Stealthy Semantic Manipulation8 Jun
  4. 2025JailBound: Jailbreaking Internal Safety Boundaries of Vision-Language Models26 May
  5. 2025Backdoor Cleaning without External Guidance in MLLM Fine-tuning22 May
  6. 2025X-Transfer Attacks: Towards Super Transferable Adversarial Attacks on CLIP8 May
  7. 2025AdPO: Enhancing the Adversarial Robustness of Large Vision-Language Models with Preference Optimization2 Apr
  8. 2025Safeguarding Vision-Language Models: Mitigating Vulnerabilities to Gaussian Noise in Perturbation-based Attacks2 Apr
  9. 2025Stealthy Backdoor Attack in Self-Supervised Learning Vision Encoders for Large Vision Language Models25 Feb
  10. 2025A Survey of Safety on Large Vision-Language Models: Attacks, Defenses and Evaluations14 Feb
  11. 2024Immune: Improving Safety Against Jailbreaks in Multi-modal LLMs via Inference-Time Alignment27 Nov
  12. 2024Steering Away from Harm: An Adaptive Approach to Defending Vision Language Model Against Jailbreaks23 Nov
34 more
  1. 2024UniGuard: Towards Universal Safety Guardrails for Jailbreak Attacks on Multimodal Large Language Models3 Nov
  2. 2024Text-Guided Attention is All You Need for Zero-Shot Robustness in Vision-Language Models29 Oct
  3. 2024BlueSuffix: Reinforced Blue Teaming for Vision-Language Models Against Jailbreak Attacks28 Oct
  4. 2024Cross-Modal Safety Mechanism Transfer in Large Vision-Language Models16 Oct
  5. 2024Break the Visual Perception: Adversarial Attacks Targeting Encoded Visual Tokens of Large Vision-Language Models9 Oct
  6. 2024ETA: Evaluating Then Aligning Safety of Vision Language Models at Inference Time9 Oct
  7. 2024AnyAttack: Towards Large-scale Self-supervised Adversarial Attacks on Vision-language Models7 Oct
  8. 2024Backdooring Vision-Language Models with Out-Of-Distribution Data2 Oct
  9. 2024TrojVLM: Backdoor Attack Against Vision Language Models28 Sep
  10. 2024Cross-modality Information Check for Detecting Jailbreaking in Multimodal Large Language Models31 Jul
  11. 2024A Survey of Attacks on Large Vision-Language Models: Resources, Advances, and Future Trends10 Jul
  12. 2024Revisiting Backdoor Attacks against Large Vision-Language Models from Domain Shift27 Jun
  13. 2024JailbreakZoo: Survey, Landscapes, and Horizons in Jailbreaking Large Language and Vision-Language Models26 Jun
  14. 2024From LLMs to MLLMs: Exploring the Landscape of Multimodal Jailbreaking21 Jun
  15. 2024SPA-VL: A Comprehensive Safety Preference Alignment Dataset for Vision Language Model17 Jun
  16. 2024Unbridled Icarus: A Survey of the Potential Perils of Image Inputs in Multimodal Large Language Model Security8 Apr
  17. 2024Eyes Closed, Safety On: Protecting Multimodal LLMs via Image-to-Text Transformation14 Mar
  18. 2024AdaShield: Safeguarding Multimodal Large Language Models from Structure-based Attack via Adaptive Shield Prompting14 Mar
  19. 2024B-AVIBench: Towards Evaluating the Robustness of Large Vision-Language Model on Black-box Adversarial Visual-Instructions14 Mar
  20. 2024ImgTrojan: Jailbreaking Vision-Language Models with ONE Image5 Mar
  21. 2024Test-Time Backdoor Attacks on Multimodal Large Language Models13 Feb
  22. 2024Shadowcast: Stealthy Data Poisoning Attacks Against Vision-Language Models5 Feb
  23. 2024Safety of Multimodal Large Language Models on Images and Texts1 Feb
  24. 2024Inducing High Energy-Latency of Large Vision-Language Models with Verbose Images20 Jan
  25. 2024Pre-trained Model Guided Fine-Tuning for Zero-Shot Adversarial Robustness9 Jan
  26. 2024MLLM-Protector: Ensuring MLLM's Safety without Hurting Performance5 Jan
  27. 2023OT-Attack: Enhancing Adversarial Transferability of Vision-Language Models via Optimal Transport Optimization7 Dec
  28. 2023BadCLIP: Trigger-Aware Prompt Learning for Backdoor Attacks on CLIP26 Nov
  29. 2023FigStep: Jailbreaking Large Vision-Language Models via Typographic Visual Prompts9 Nov
  30. 2023Jailbreak in pieces: Compositional Adversarial Attacks on Multi-Modal Language Models26 Jul
  31. 2023Set-level Guidance Attack: Boosting Adversarial Transferability of Vision-Language Pre-training Models26 Jul
  32. 2023Visual Adversarial Examples Jailbreak Aligned Large Language Models22 Jun
  33. 2023On Evaluating Adversarial Robustness of Large Vision-Language Models26 May
  34. 2022Understanding Zero-Shot Adversarial Robustness for Large-Scale Models14 Dec

Other references · 98

  1. 2020Language Models are Few-Shot Learnerscited by 32,419
  2. 2022Training language models to follow instructions with human feedbackcited by 15,971
  3. 2019Exploring the Limits of Transfer Learning with a Unified Text-to-Text Transformercited by 14,599
  4. 2022Chain-of-Thought Prompting Elicits Reasoning in Large Language Modelscited by 12,459
  5. 2020Retrieval-Augmented Generation for Knowledge-Intensive NLP Taskscited by 9,297
  6. 2023Visual Instruction Tuningcited by 7,869
  7. 2014Explaining and Harnessing Adversarial Examplescited by 7,364
  8. 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,415
  9. 2023BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Modelscited by 6,137
  10. 2022Flamingo: a Visual Language Model for Few-Shot Learningcited by 4,386
  11. 2022BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generationcited by 4,322
  12. 2023Universal and Transferable Adversarial Attacks on Aligned Language Modelscited by 3,104
38 more
  1. 2016Towards Evaluating the Robustness of Neural Networkscited by 2,998
  2. 2019Fine-Tuning Language Models from Human Preferencescited by 2,150
  3. 2020REALM: Retrieval-Augmented Language Model Pre-Trainingcited by 2,141
  4. 2023A Survey on Hallucination in Large Language Models: Principles, Taxonomy, Challenges, and Open Questionscited by 1,601
  5. 2022Least-to-Most Prompting Enables Complex Reasoning in Large Language Modelscited by 1,407
  6. 2019Certified Adversarial Robustness via Randomized Smoothingcited by 1,340
  7. 2020Leveraging Passage Retrieval with Generative Models for Open Domain Question Answeringcited by 1,210
  8. 2023Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!cited by 1,154
  9. 2017Targeted Backdoor Attacks on Deep Learning Systems Using Data Poisoningcited by 1,055
  10. 2017Multimodal Machine Learning: A Survey and Taxonomycited by 1,013
  11. 2023A Survey on Multimodal Large Language Modelscited by 909
  12. 2023AutoDAN: Generating Stealthy Jailbreak Prompts on Aligned Large Language Modelscited by 785
  13. 2023The RefinedWeb Dataset for Falcon LLM: Outperforming Curated Corpora with Web Data, and Web Data Onlycited by 763
  14. 2019Universal Adversarial Triggers for Attacking and Analyzing NLPcited by 722
  15. 2021Deduplicating Training Data Makes Language Models Bettercited by 714
  16. 2017ZOO: Zeroth Order Optimization based Black-box Attacks to Deep Neural Networks without Training Substitute Modelscited by 666
  17. 2022PAL: Program-aided Language Modelscited by 652
  18. 2023Baseline Defenses for Adversarial Attacks Against Aligned Language Modelscited by 651
  19. 2023HallusionBench: An Advanced Diagnostic Suite for Entangled Language Hallucination and Visual Illusion in Large Vision-Language Modelscited by 583
  20. 2018Fine-Pruning: Defending Against Backdooring Attacks on Deep Neural Networkscited by 580
  21. 2023mPLUG-Owl2: Revolutionizing Multi-modal Large Language Model with Modality Collaborationcited by 547
  22. 2018Spectral Signatures in Backdoor Attackscited by 484
  23. 2019STRIP: A Defence Against Trojan Attacks on Deep Neural Networkscited by 459
  24. 2023Mitigating Hallucination in Large Multi-Modal Models via Robust Instruction Tuningcited by 445
  25. 2023Detecting Language Model Attacks with Perplexitycited by 431
  26. 2023Jailbreak and Guard Aligned Language Models with Only Few In-Context Demonstrationscited by 431
  27. 2023GPT-4 Is Too Smart To Be Safe: Stealthy Chat with LLMs via Ciphercited by 423
  28. 2022Multimodal Learning with Transformers: A Surveycited by 400
  29. 2023Safety-Tuned LLaMAs: Lessons From Improving the Safety of Large Language Models that Follow Instructionscited by 387
  30. 2019Hidden Trigger Backdoor Attackscited by 371
  31. 2023DeepInception: Hypnotize Large Language Model to Be Jailbreakercited by 367
  32. 2024Prismatic VLMs: Investigating the Design Space of Visually-Conditioned Language Modelscited by 339
  33. 2021Neural Attention Distillation: Erasing Backdoor Triggers from Deep Neural Networkscited by 324
  34. 2023Principle-Driven Self-Alignment of Language Models from Scratch with Minimal Human Supervisioncited by 318
  35. 2020Input-Aware Dynamic Backdoor Attackcited by 313
  36. 2023Shadow Alignment: The Ease of Subverting Safely-Aligned Language Modelscited by 297
  37. 2021Anti-Backdoor Learning: Training Clean Models on Poisoned Datacited by 263
  38. 2022Making Large Language Models Better Reasoners with Step-Aware Verifiercited by 233

50 most cited of 98 All 98 (JSON)

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2025Do We Really Need Curated Malicious Data for Safety Alignment in Multi-modal Large Language Models?
  2. 2025SafeMLRM: Demystifying Safety in Multi-modal Large Reasoning Models
  3. 2024Cross-Modal Safety Alignment: Is textual unlearning all you need?
  4. 2025DREAM: Disentangling Risks to Enhance Safety Alignment in Multimodal Large Language Models
  5. 2025SafeMT: Multi-turn Safety for Multimodal Language Models
  6. 2025When Safe Unimodal Inputs Collide: Optimizing Reasoning Chains for Cross-Modal Safety in Multimodal Large Language Models
  7. 2026SafeSteer: A Decoding-level Defense Mechanism for Multimodal Large Language Models
  8. 2026MME-Safety: A Fine-grained Benchmark for Safety Evaluation of MLLMs