# Innocence in the Crossfire: Roles of Skip Connections in Jailbreaking Visual Language Models

arXiv: https://arxiv.org/abs/2507.13761 · HTML: https://mlsanity.com/paper/2507.13761/

## 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 46 of its 56 references that are in the map.
- First posted on arXiv 18 Jul 2025.

## 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, 30 of 46)

- [Attention Is All You Need](https://mlsanity.com/paper/1706.03762/) (2017)
- [GPT-4 Technical Report](https://mlsanity.com/paper/2303.08774/) (2023)
- [Visual Instruction Tuning](https://mlsanity.com/paper/2304.08485/) (2023)
- [Layer Normalization](https://mlsanity.com/paper/1607.06450/) (2016)
- [Improved Baselines with Visual Instruction Tuning](https://mlsanity.com/paper/2310.03744/) (2023)
- [Universal and Transferable Adversarial Attacks on Aligned Language Models](https://mlsanity.com/paper/2307.15043/) (2023)
- [MiniGPT-4: Enhancing Vision-Language Understanding with Advanced Large Language Models](https://mlsanity.com/paper/2304.10592/) (2023)
- [Qwen-VL: A Versatile Vision-Language Model for Understanding, Localization, Text Reading, and Beyond](https://mlsanity.com/paper/2308.12966/) (2023)
- [ViLBERT: Pretraining Task-Agnostic Visiolinguistic Representations for Vision-and-Language Tasks](https://mlsanity.com/paper/1908.02265/) (2019)
- [Jailbroken: How Does LLM Safety Training Fail?](https://mlsanity.com/paper/2307.02483/) (2023)
- [Fine-tuning Aligned Language Models Compromises Safety, Even When Users Do Not Intend To!](https://mlsanity.com/paper/2310.03693/) (2023)
- [Refusal in Language Models Is Mediated by a Single Direction](https://mlsanity.com/paper/2406.11717/) (2024)
- [Transformers in Vision: A Survey](https://mlsanity.com/paper/2101.01169/) (2021)
- [Chameleon: Mixed-Modal Early-Fusion Foundation Models](https://mlsanity.com/paper/2405.09818/) (2024)
- [BeaverTails: Towards Improved Safety Alignment of LLM via a Human-Preference Dataset](https://mlsanity.com/paper/2307.04657/) (2023)
- [Vision-Language Models for Vision Tasks: A Survey](https://mlsanity.com/paper/2304.00685/) (2023)
- [Emu3: Next-Token Prediction is All You Need](https://mlsanity.com/paper/2409.18869/) (2024)
- [MiniCPM: Unveiling the Potential of Small Language Models with Scalable Training Strategies](https://mlsanity.com/paper/2404.06395/) (2024)
- [The Hateful Memes Challenge: Detecting Hate Speech in Multimodal Memes](https://mlsanity.com/paper/2005.04790/) (2020)
- [Jailbreak and Guard Aligned Language Models with Only Few In-Context Demonstrations](https://mlsanity.com/paper/2310.06387/) (2023)
- [DeepInception: Hypnotize Large Language Model to Be Jailbreaker](https://mlsanity.com/paper/2311.03191/) (2023)
- [Are aligned neural networks adversarially aligned?](https://mlsanity.com/paper/2306.15447/) (2023)
- [Shadow Alignment: The Ease of Subverting Safely-Aligned Language Models](https://mlsanity.com/paper/2310.02949/) (2023)
- [Jailbreak in pieces: Compositional Adversarial Attacks on Multi-Modal Language Models](https://mlsanity.com/paper/2307.14539/) (2023)
- [A Wolf in Sheep's Clothing: Generalized Nested Jailbreak Prompts can Fool Large Language Models Easily](https://mlsanity.com/paper/2311.08268/) (2023)
- [LoRA Fine-tuning Efficiently Undoes Safety Training in Llama 2-Chat 70B](https://mlsanity.com/paper/2310.20624/) (2023)
- [Automatically Auditing Large Language Models via Discrete Optimization](https://mlsanity.com/paper/2303.04381/) (2023)
- [A Survey of Vision-Language Pre-Trained Models](https://mlsanity.com/paper/2202.10936/) (2022)
- [ProSA: Assessing and Understanding the Prompt Sensitivity of LLMs](https://mlsanity.com/paper/2410.12405/) (2024)
- [AutoDAN: Interpretable Gradient-Based Adversarial Attacks on Large Language Models](https://mlsanity.com/paper/2310.15140/) (2023)

## Abstract (arXiv)

Language models are highly sensitive to prompt formulations - small changes in input can drastically alter their output. This raises a critical question: To what extent can prompt sensitivity be exploited to generate inapt content? In this paper, we investigate how discrete components of prompt design influence the generation of inappropriate content in Visual Language Models (VLMs). Specifically, we analyze the impact of three key factors on successful jailbreaks: (a) the inclusion of detailed visual information, (b) the presence of adversarial examples, and (c) the use of positively framed beginning phrases. Our findings reveal that while a VLM can reliably distinguish between benign and harmful inputs in unimodal settings (text-only or image-only), this ability significantly degrades in multimodal contexts. Each of the three factors is independently capable of triggering a jailbreak, and we show that even a small number of in-context examples (as few as three) can push the model toward generating inappropriate outputs. Furthermore, we propose a framework that utilizes a skip-connection between two internal layers of the VLM, which substantially increases jailbreak success rates, even when using benign images. Finally, we demonstrate that memes, often perceived as humorous or harmless, can be as effective as toxic visuals in eliciting harmful content, underscoring the subtle and complex vulnerabilities of VLMs.