paper-with-me

홈 › Papers

Simulated Ensemble Attack: Transferring Jailbreaks Across Fine-tuned Vision-Language Models

2025-08-03 · Ruofan Wang, Xin Wang, Yang Yao, Juncheng Li, Xuan Tong, Xingjun Ma arxiv

The widespread practice of fine-tuning open-source Vision-Language Models (VLMs) raises a critical security concern: jailbreak vulnerabilities in base models may persist in downstream variants, enabling transferable attacks across fine-tuned systems. To investigate this risk, we propose the Simulated Ensemble Attack (SEA), a grey-box jailbreak framework that assumes full access to the base VLM but no knowledge of the fine-tuned target. SEA enhances transferability via Fine-tuning Trajectory Simulation (FTS), which models bounded parameter variations in the vision encoder, and Targeted Prompt Guidance (TPG), which stabilizes adversarial optimization through auxiliary textual guidance. Experiments on the Qwen2-VL family demonstrate that SEA achieves consistently high transfer success and toxicity rates across diverse fine-tuned variants, including safety-enhanced models, while standard PGD-based image jailbreaks exhibit negligible transferability. Further analysis reveals that fine-tuning primarily induces localized parameter shifts around the base model, explaining why attacks optimized over a simulated neighborhood transfer effectively. We also show that SEA generalizes across different base generations (e.g., Qwen2.5/3-VL), indicating that its effectiveness arises from shared fine-tuning-induced behaviors rather than architecture- or initialization-specific factors.

📄 PDF Abstract BibTeX arXiv:2508.01741

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Failures to Find Transferable Image Jailbreaks Between Vision-Language Models

2024-07-21 · Rylan Schaeffer, Dan Valentine, Luke Bailey, James Chua 외

The integration of new modalities into frontier AI systems offers exciting capabilities, but also increases the possibility such systems can be adversarially manipulated in undesirable ways. In this work, we focus on a p…

Instruction FollowingLanguage ModellingSafety Alignment

EnJa: Ensemble Jailbreak on Large Language Models

2024-08-07 · Jiahao Zhang, Zilong Wang, Ruofan Wang, Xingjun Ma 외

As Large Language Models (LLMs) are increasingly being deployed in safety-critical applications, their vulnerability to potential jailbreaks -- malicious prompts that can disable the safety mechanism of LLMs -- has attra…

Safety Alignment

LLM Defenses Are Not Robust to Multi-Turn Human Jailbreaks Yet

2024-08-27 · Nathaniel Li, Ziwen Han, Ian Steneker, Willow Primack 외

Recent large language model (LLM) defenses have greatly improved models' ability to refuse harmful queries, even when adversarially attacked. However, LLM defenses are primarily evaluated against automated adversarial at…

Language ModelingLanguage ModellingLarge Language ModelMachine Unlearning+1

"I am bad": Interpreting Stealthy, Universal and Robust Audio Jailbreaks in Audio-Language Models

2025-02-02 · Isha Gupta, David Khachaturov, Robert Mullins

The rise of multimodal large language models has introduced innovative human-machine interaction paradigms but also significant challenges in machine learning safety. Audio-Language Models (ALMs) are especially relevant …

An Adaptive Model Ensemble Adversarial Attack for Boosting Adversarial Transferability

2023-08-05 · ICCV 2023 1 · Bin Chen, Jia-Li Yin, Shukai Chen, Bo-Hao Chen 외

While the transferability property of adversarial examples allows the adversary to perform black-box attacks (i.e., the attacker has no knowledge about the target model), the transfer-based adversarial attacks have gaine…

Adversarial Attack