Two Birds, One Projection: Harmonizing Safety and Utility in LVLMs via Inference-Time Feature Projection
Authors:
- 12 September 2026
- book chapter
- Published by Springer Nature
- p. 319-336
Abstract
No abstract availableKeywords
All Related Versions
This publication has 21 references indexed in Scilit:
- Jailbreaking Multimodal Large Language Models via Shuffle Inconsistency2025
- Nullu: Mitigating Object Hallucinations in Large Vision-Language Models via HalluSpace Projection2025
- Immune: Improving Safety Against Jailbreaks in Multi-modal LLMs via Inference-Time Alignment2025
- Steering Away from Harm: An Adaptive Approach to Defending Vision Language Model Against Jailbreaks2025
- Images are Achilles’ Heel of Alignment: Exploiting Visual Vulnerabilities for Jailbreaking Multimodal Large Language Models2024
- AdaShield : Safeguarding Multimodal Large Language Models from Structure-Based Attack via Adaptive Shield Prompting2024
- Eyes Closed, Safety on: Protecting Multimodal LLMs via Image-to-Text Transformation2024
- MM-SafetyBench: A Benchmark for Safety Evaluation of Multimodal Large Language Models2024
- Unveiling Vulnerabilities in Large Vision-Language Models: The SAVJ Jailbreak Approach2024
- Intern VL: Scaling up Vision Foundation Models and Aligning for Generic Visual-Linguistic Tasks2024