RapGuard: Safeguarding Multimodal Large Language Models via Rationale-aware Defensive Prompting
Fuente:
arXiv
Saved in:
| Main Authors: | , , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866915080751284224 |
|---|---|
| author | Jiang, Yilei Tan, Yingshui Yue, Xiangyu |
| author_facet | Jiang, Yilei Tan, Yingshui Yue, Xiangyu |
| contents | While Multimodal Large Language Models (MLLMs) have made remarkable progress in vision-language reasoning, they are also more susceptible to producing harmful content compared to models that focus solely on text. Existing defensive prompting techniques rely on a static, unified safety guideline that fails to account for the specific risks inherent in different multimodal contexts. To address these limitations, we propose RapGuard, a novel framework that uses multimodal chain-of-thought reasoning to dynamically generate scenario-specific safety prompts. RapGuard enhances safety by adapting its prompts to the unique risks of each input, effectively mitigating harmful outputs while maintaining high performance on benign tasks. Our experimental results across multiple MLLM benchmarks demonstrate that RapGuard achieves state-of-the-art safety performance, significantly reducing harmful content without degrading the quality of responses. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_18826 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | RapGuard: Safeguarding Multimodal Large Language Models via Rationale-aware Defensive Prompting Jiang, Yilei Tan, Yingshui Yue, Xiangyu Computation and Language While Multimodal Large Language Models (MLLMs) have made remarkable progress in vision-language reasoning, they are also more susceptible to producing harmful content compared to models that focus solely on text. Existing defensive prompting techniques rely on a static, unified safety guideline that fails to account for the specific risks inherent in different multimodal contexts. To address these limitations, we propose RapGuard, a novel framework that uses multimodal chain-of-thought reasoning to dynamically generate scenario-specific safety prompts. RapGuard enhances safety by adapting its prompts to the unique risks of each input, effectively mitigating harmful outputs while maintaining high performance on benign tasks. Our experimental results across multiple MLLM benchmarks demonstrate that RapGuard achieves state-of-the-art safety performance, significantly reducing harmful content without degrading the quality of responses. |
| title | RapGuard: Safeguarding Multimodal Large Language Models via Rationale-aware Defensive Prompting |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2412.18826 |