Two Birds, One Projection: Harmonizing Safety and Utility in LVLMs via Inference-time Feature Projection

Existing jailbreak defence frameworks for Large Vision-Language Models often suffer from a safety utility tradeoff, where strengthening safety inadvertently degrades performance on general visual-grounded reasoning tasks. In this work, we investigate whether safety and utility are inherently antagonistic objectives. We focus on a modality induced bias direction consistently observed across datasets, which arises from suboptimal coupling between the Large Language Model backbone and visual encoders. We further demonstrate that this direction undermines performance on both tasks. Leveraging this insight, we propose Two Birds, One Projection, an efficient inference time jailbreak defence that projects cross-modal features onto the null space of the identified bias direction to remove the corresponding components. Requiring only a single forward pass, our method effectively breaks the conventional tradeoff, simultaneously improving both safety and utility across diverse benchmarks.

Key Contributions

• Identifies a modality-induced bias direction in LVLMs arising from suboptimal cross-modal coupling that degrades both safety and utility
• Proposes TBOP, an efficient single-forward-pass inference-time defense via null-space projection
• Breaks the conventional safety-utility tradeoff, improving both jailbreak defense and general visual reasoning performance

🛡️ Threat Analysis

Details

Similar Papers