# BaThe: Defense against the Jailbreak Attack in Multimodal Large Language Models by Treating Harmful Instruction as Backdoor Trigger

arXiv: https://arxiv.org/abs/2408.09093 · HTML: https://mlsanity.com/paper/2408.09093/

## Key facts

- Cited by 8 papers in the map (data through 14 Oct 2026).
- Placed in Vision-language and text-to-image safety, one of 12 directions in LLM safety and security.
- Builds on 18 of its 37 references that are in the map.
- First posted on arXiv 17 Aug 2024.

## Placed in

- 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, 18 of 18)

- [Adam: A Method for Stochastic Optimization](https://mlsanity.com/paper/1412.6980/) (2014)
- [PyTorch: An Imperative Style, High-Performance Deep Learning Library](https://mlsanity.com/paper/1912.01703/) (2019)
- [Training language models to follow instructions with human feedback](https://mlsanity.com/paper/2203.02155/) (2022)
- [Universal and Transferable Adversarial Attacks on Aligned Language Models](https://mlsanity.com/paper/2307.15043/) (2023)
- [MMBench: Is Your Multi-modal Model an All-around Player?](https://mlsanity.com/paper/2307.06281/) (2023)
- [Ignore Previous Prompt: Attack Techniques For Language Models](https://mlsanity.com/paper/2211.09527/) (2022)
- [Safe RLHF: Safe Reinforcement Learning from Human Feedback](https://mlsanity.com/paper/2310.12773/) (2023)
- [Prompt Injection attack against LLM-integrated Applications](https://mlsanity.com/paper/2306.05499/) (2023)
- [FigStep: Jailbreaking Large Vision-Language Models via Typographic Visual Prompts](https://mlsanity.com/paper/2311.05608/) (2023)
- [Backdooring Instruction-Tuned Large Language Models with Virtual Prompt Injection](https://mlsanity.com/paper/2307.16888/) (2023)
- [On the Adversarial Robustness of Multi-Modal Foundation Models](https://mlsanity.com/paper/2308.10741/) (2023)
- [AdaShield: Safeguarding Multimodal Large Language Models from Structure-based Attack via Adaptive Shield Prompting](https://mlsanity.com/paper/2403.09513/) (2024)
- [Eyes Closed, Safety On: Protecting Multimodal LLMs via Image-to-Text Transformation](https://mlsanity.com/paper/2403.09572/) (2024)
- [MLLM-Protector: Ensuring MLLM's Safety without Hurting Performance](https://mlsanity.com/paper/2401.02906/) (2024)
- [How Many Unicorns Are in This Image? A Safety Evaluation Benchmark for Vision LLMs](https://mlsanity.com/paper/2311.16101/) (2023)
- [InferAligner: Inference-Time Alignment for Harmlessness through Cross-Model Guidance](https://mlsanity.com/paper/2401.11206/) (2024)
- [ImgTrojan: Jailbreaking Vision-Language Models with ONE Image](https://mlsanity.com/paper/2403.02910/) (2024)
- [Gaining Wisdom from Setbacks: Aligning Large Language Models via Mistake Analysis](https://mlsanity.com/paper/2310.10477/) (2023)

## Built on by (most cited first, 8 of 8)

- [A Survey of Safety on Large Vision-Language Models: Attacks, Defenses and Evaluations](https://mlsanity.com/paper/2502.14881/) (2025)
- [Distraction is All You Need for Multimodal Large Language Model Jailbreaking](https://mlsanity.com/paper/2502.10794/) (2025)
- [MCIP: Protecting MCP Safety via Model Contextual Integrity Protocol](https://mlsanity.com/paper/2505.14590/) (2025)
- [Jailbreak Attacks and Defenses against Multimodal Generative Models: A Survey](https://mlsanity.com/paper/2411.09259/) (2024)
- [Exploring Response Uncertainty in MLLMs: An Empirical Evaluation under Misleading Scenarios](https://mlsanity.com/paper/2411.02708/) (2024)
- [Robust Anti-Backdoor Instruction Tuning in LVLMs](https://mlsanity.com/paper/2506.05401/) (2025)
- [SmoothGuard: Defending Multimodal Large Language Models with Noise Perturbation and Clustering Aggregation](https://mlsanity.com/paper/2510.26830/) (2025)
- [COMIC: Reference-Aware Safety Gating for Multimodal Large Language Models](https://mlsanity.com/paper/2608.17234/) (2026)

## Abstract (arXiv)

Multimodal Large Language Models (MLLMs) have showcased impressive performance in a variety of multimodal tasks. On the other hand, the integration of additional image modality may allow the malicious users to inject harmful content inside the images for jailbreaking. Unlike text-based LLMs, where adversaries need to select discrete tokens to conceal their malicious intent using specific algorithms, the continuous nature of image signals provides a direct opportunity for adversaries to inject harmful intentions. In this work, we propose BaThe (Backdoor Trigger Shield), a simple yet effective jailbreak defense mechanism. Our work is motivated by recent research on jailbreak backdoor attack and virtual prompt backdoor attack in generative language models. Jailbreak backdoor attack uses harmful instructions combined with manually crafted strings as triggers to make the backdoored model generate prohibited responses. We assume that harmful instructions can function as triggers, and if we alternatively set rejection responses as the triggered response, the backdoored model then can defend against jailbreak attacks. We achieve this by utilizing virtual rejection prompt, similar to the virtual prompt backdoor attack. We embed the virtual rejection prompt into the soft text embeddings, which we call ``wedge''. Our comprehensive experiments demonstrate that BaThe effectively mitigates various types of jailbreak attacks and is adaptable to defend against unseen attacks, with minimal impact on MLLMs' performance.