# One Perturbation, Two Failure Modes: Probing VLM Safety via Embedding-Guided Typographic Perturbations

arXiv: https://arxiv.org/abs/2604.25102 · HTML: https://mlsanity.com/paper/2604.25102/

## 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 10 of its 10 references that are in the map.
- First posted on arXiv 28 Apr 2026.

## Placed in

- Multimodal LLMs / vision-language models / [MLLM jailbreaks and adversarial robustness](https://mlsanity.com/mllm/mllm-jailbreaks-and-adversarial-robustness/)
- LLM safety and security / [Vision-language and text-to-image safety](https://mlsanity.com/safety/vision-language-and-text-to-image-safety/)

## Builds on (most cited first, 10 of 10)

- [FigStep: Jailbreaking Large Vision-Language Models via Typographic Visual Prompts](https://mlsanity.com/paper/2311.05608/) (2023)
- [SALAD-Bench: A Hierarchical and Comprehensive Safety Benchmark for Large Language Models](https://mlsanity.com/paper/2402.05044/) (2024)
- [How Robust is Google's Bard to Adversarial Image Attacks?](https://mlsanity.com/paper/2309.11751/) (2023)
- [AdvCLIP: Downstream-agnostic Adversarial Examples in Multimodal Contrastive Learning](https://mlsanity.com/paper/2308.07026/) (2023)
- [Rethinking Model Ensemble in Transfer-based Adversarial Attacks](https://mlsanity.com/paper/2303.09105/) (2023)
- [Jina CLIP: Your CLIP Model Is Also Your Text Retriever](https://mlsanity.com/paper/2405.20204/) (2024)
- [Defense-Prefix for Preventing Typographic Attacks on CLIP](https://mlsanity.com/paper/2304.04512/) (2023)
- [SceneTAP: Scene-Coherent Typographic Adversarial Planner against Vision-Language Models in Real-World Environments](https://mlsanity.com/paper/2412.00114/) (2024)
- [SCAM: A Real-World Typographic Robustness Evaluation for Multimodal Foundation Models](https://mlsanity.com/paper/2504.04893/) (2025)
- [Typographic Attacks in a Multi-Image Setting](https://mlsanity.com/paper/2502.08193/) (2025)

## Abstract (arXiv)

Typographic prompt injection exploits vision language models' (VLMs) ability to read text rendered in images, posing a growing threat as VLMs power autonomous agents. Prior work typically focus on maximizing attack success rate (ASR) but does not explain why certain renderings bypass safety alignment. We make two contributions. First, an empirical study across four VLMs including GPT-4o and Claude, twelve font sizes, and ten transformations reveals that multimodal embedding distance strongly predicts ASR (r=-0.71 to -0.93, p<0.01), providing an interpretable, model agnostic proxy. Since embedding distance predicts ASR, reducing it should improve attack success, but the relationship is mediated by two factors: perceptual readability (whether the VLM can parse the text) and safety alignment (whether it refuses to comply). Second, we use this as a red teaming tool: we directly maximize image text embedding similarity under bounded ℓ_∞ perturbations via CWA-SSA across four surrogate embedding models, stress testing both factors without access to the target model. Experiments across five degradation settings on GPT-4o, Claude Sonnet 4.5, Mistral-Large-3, and Qwen3-VL confirm that optimization recovers readability and reduces safety aligned refusals as two co-occurring effects, with the dominant mechanism depending on the model's safety filter strength and the degree of visual degradation.