Safety Alignment for Vision Language Models
This addresses safety risks in VLMs for users deploying multimodal AI systems, though it is incremental as it builds upon existing models like LLaVA-v1.5.
The paper tackles the vulnerability of Vision Language Models (VLMs) to attacks through visual modality features by enhancing safety alignment with added safety modules and a two-stage training process, achieving a safety score of 8.26 on the RTVLM benchmark, surpassing GPT-4V.
Benefiting from the powerful capabilities of Large Language Models (LLMs), pre-trained visual encoder models connected to an LLMs can realize Vision Language Models (VLMs). However, existing research shows that the visual modality of VLMs is vulnerable, with attackers easily bypassing LLMs' safety alignment through visual modality features to launch attacks. To address this issue, we enhance the existing VLMs' visual modality safety alignment by adding safety modules, including a safety projector, safety tokens, and a safety head, through a two-stage training process, effectively improving the model's defense against risky images. For example, building upon the LLaVA-v1.5 model, we achieve a safety score of 8.26, surpassing the GPT-4V on the Red Teaming Visual Language Models (RTVLM) benchmark. Our method boasts ease of use, high flexibility, and strong controllability, and it enhances safety while having minimal impact on the model's general performance. Moreover, our alignment strategy also uncovers some possible risky content within commonly used open-source multimodal datasets. Our code will be open sourced after the anonymous review.