Immune: Improving Safety Against Jailbreaks in Multi-modal LLMs via Inference-Time Alignment
Fuente:
arXiv
Salvato in:
| Autori principali: | , , , , , , , , , |
|---|---|
| Natura: | Preprint |
| Pubblicazione: |
2024
|
| Soggetti: | |
| Accesso online: | |
| Tags: |
Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
|
| _version_ | 1866915342519894016 |
|---|---|
| author | Ghosal, Soumya Suvra Chakraborty, Souradip Singh, Vaibhav Guan, Tianrui Wang, Mengdi Velasquez, Alvaro Beirami, Ahmad Huang, Furong Manocha, Dinesh Bedi, Amrit Singh |
| author_facet | Ghosal, Soumya Suvra Chakraborty, Souradip Singh, Vaibhav Guan, Tianrui Wang, Mengdi Velasquez, Alvaro Beirami, Ahmad Huang, Furong Manocha, Dinesh Bedi, Amrit Singh |
| contents | With the widespread deployment of Multimodal Large Language Models (MLLMs) for visual-reasoning tasks, improving their safety has become crucial. Recent research indicates that despite training-time safety alignment, these models remain vulnerable to jailbreak attacks. In this work, we first highlight an important safety gap to describe that alignment achieved solely through safety training may be insufficient against jailbreak attacks. To address this vulnerability, we propose Immune, an inference-time defense framework that leverages a safe reward model through controlled decoding to defend against jailbreak attacks. Additionally, we provide a mathematical characterization of Immune, offering insights on why it improves safety against jailbreaks. Extensive evaluations on diverse jailbreak benchmarks using recent MLLMs reveal that Immune effectively enhances model safety while preserving the model's original capabilities. For instance, against text-based jailbreak attacks on LLaVA-1.6, Immune reduces the attack success rate by 57.82% and 16.78% compared to the base MLLM and state-of-the-art defense strategy, respectively. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2411_18688 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Immune: Improving Safety Against Jailbreaks in Multi-modal LLMs via Inference-Time Alignment Ghosal, Soumya Suvra Chakraborty, Souradip Singh, Vaibhav Guan, Tianrui Wang, Mengdi Velasquez, Alvaro Beirami, Ahmad Huang, Furong Manocha, Dinesh Bedi, Amrit Singh Cryptography and Security Artificial Intelligence Machine Learning With the widespread deployment of Multimodal Large Language Models (MLLMs) for visual-reasoning tasks, improving their safety has become crucial. Recent research indicates that despite training-time safety alignment, these models remain vulnerable to jailbreak attacks. In this work, we first highlight an important safety gap to describe that alignment achieved solely through safety training may be insufficient against jailbreak attacks. To address this vulnerability, we propose Immune, an inference-time defense framework that leverages a safe reward model through controlled decoding to defend against jailbreak attacks. Additionally, we provide a mathematical characterization of Immune, offering insights on why it improves safety against jailbreaks. Extensive evaluations on diverse jailbreak benchmarks using recent MLLMs reveal that Immune effectively enhances model safety while preserving the model's original capabilities. For instance, against text-based jailbreak attacks on LLaVA-1.6, Immune reduces the attack success rate by 57.82% and 16.78% compared to the base MLLM and state-of-the-art defense strategy, respectively. |
| title | Immune: Improving Safety Against Jailbreaks in Multi-modal LLMs via Inference-Time Alignment |
| topic | Cryptography and Security Artificial Intelligence Machine Learning |
| url | https://arxiv.org/abs/2411.18688 |