Imagen de Portada

Rethinking Bottlenecks in Safety Fine-Tuning of Vision Language Models

Fuente: arXiv

Guardado en:

Detalles Bibliográficos
Autores principales:	Ding, Yi, Li, Lijun, Cao, Bing, Shao, Jing
Formato:	Preprint
Publicado:	2025
Materias:	Computer Vision and Pattern Recognition Computation and Language Cryptography and Security
Acceso en línea:	Acceder al recurso
Etiquetas:	Agregar Etiqueta Sin Etiquetas, Sea el primero en etiquetar este registro!

Ejemplares similares

Visual Contextual Attack: Jailbreaking MLLMs with Image-Driven Context Injection
por: Miao, Ziqi, et al.
Publicado: (2025)

Self-adaptive Dataset Construction for Real-World Multimodal Safety Scenarios
por: Qu, Jingen, et al.
Publicado: (2025)

Contextual Image Attack: How Visual Context Exposes Multimodal Safety Vulnerabilities
por: Xiong, Yuan, et al.
Publicado: (2025)

Image-Based Geolocation Using Large Vision-Language Models
por: Liu, Yi, et al.
Publicado: (2024)

Lazy Layers to Make Fine-Tuned Diffusion Models More Traceable
por: Liu, Haozhe, et al.
Publicado: (2024)

To See is Not to Learn: Protecting Multimodal Data from Unauthorized Fine-Tuning of Large Vision-Language Model
por: Zhao, Chengshuai, et al.
Publicado: (2026)

VLSBench: Unveiling Visual Leakage in Multimodal Safety
por: Hu, Xuhao, et al.
Publicado: (2024)

Recovering the Pre-Fine-Tuning Weights of Generative Models
por: Horwitz, Eliahu, et al.
Publicado: (2024)

Are Vision-Language Models Safe in the Wild? A Meme-Based Benchmark Study
por: Lee, DongGeon, et al.
Publicado: (2025)

Effective and Efficient Adversarial Detection for Vision-Language Models via A Single Vector
por: Huang, Youcheng, et al.
Publicado: (2024)

Evolving Contextual Safety in Multi-Modal Large Language Models via Inference-Time Self-Reflective Memory
por: Zhang, Ce, et al.
Publicado: (2026)

DAVSP: Safety Alignment for Large Vision-Language Models via Deep Aligned Visual Safety Prompt
por: Zhang, Yitong, et al.
Publicado: (2025)

Doubly-Universal Adversarial Perturbations: Deceiving Vision-Language Models Across Both Images and Text with a Single Perturbation
por: Kim, Hee-Seon, et al.
Publicado: (2024)

PII-VisBench: Evaluating Personally Identifiable Information Safety in Vision Language Models Along a Continuum of Visibility
por: Shahariar, G M, et al.
Publicado: (2026)

DeepSight: An All-in-One LM Safety Toolkit
por: Zhang, Bo, et al.
Publicado: (2026)

R-TPT: Improving Adversarial Robustness of Vision-Language Models through Test-Time Prompt Tuning
por: Sheng, Lijun, et al.
Publicado: (2025)

The Safety Reminder: A Soft Prompt to Reactivate Delayed Safety Awareness in Vision-Language Models
por: Tang, Peiyuan, et al.
Publicado: (2025)

A Survey of Safety on Large Vision-Language Models: Attacks, Defenses and Evaluations
por: Ye, Mang, et al.
Publicado: (2025)

JailbreakZoo: Survey, Landscapes, and Horizons in Jailbreaking Large Language and Vision-Language Models
por: Jin, Haibo, et al.
Publicado: (2024)

Safety at Scale: A Comprehensive Survey of Large Model and Agent Safety
por: Ma, Xingjun, et al.
Publicado: (2025)

MLLM-Protector: Ensuring MLLM's Safety without Hurting Performance
por: Pi, Renjie, et al.
Publicado: (2024)

Rethinking Machine Unlearning in Image Generation Models
por: Liu, Renyang, et al.
Publicado: (2025)

X-Boundary: Establishing Exact Safety Boundary to Shield LLMs from Multi-Turn Jailbreaks without Compromising Usability
por: Lu, Xiaoya, et al.
Publicado: (2025)

Few-Shot Adversarial Prompt Learning on Vision-Language Models
por: Zhou, Yiwei, et al.
Publicado: (2024)

Are GUI Agents Focused Enough? Automated Distraction via Semantic-level UI Element Injection
por: Yang, Wenkui, et al.
Publicado: (2026)

VERA-V: Variational Inference Framework for Jailbreaking Vision-Language Models
por: Liao, Qilin, et al.
Publicado: (2025)

CAT: Concept-level backdoor ATtacks for Concept Bottleneck Models
por: Lai, Songning, et al.
Publicado: (2024)

Jailbreaking Safeguarded Text-to-Image Models via Large Language Models
por: Jiang, Zhengyuan, et al.
Publicado: (2025)

Spot Risks Before Speaking! Unraveling Safety Attention Heads in Large Vision-Language Models
por: Zheng, Ziwei, et al.
Publicado: (2025)

Revisiting Data Auditing in Large Vision-Language Models
por: Zhu, Hongyu, et al.
Publicado: (2025)

Backdoor Mitigation in Object Detection via Adversarial Fine-Tuning
por: Dunnett, Kealan, et al.
Publicado: (2026)

Differentially Private Bias-Term Fine-tuning of Foundation Models
por: Bu, Zhiqi, et al.
Publicado: (2022)

Differentially Private Fine-Tuning of Diffusion Models
por: Tsai, Yu-Lin, et al.
Publicado: (2024)

Responsible Diffusion: A Comprehensive Survey on Safety, Ethics, and Trust in Diffusion Models
por: Wei, Kang, et al.
Publicado: (2025)

Rethinking Robust Adversarial Concept Erasure in Diffusion Models
por: Yin, Qinghong, et al.
Publicado: (2025)

Guarding the Gate: ConceptGuard Battles Concept-Level Backdoors in Concept Bottleneck Models
por: Lai, Songning, et al.
Publicado: (2024)

Membership Inference Attacks against Large Vision-Language Models
por: Li, Zhan, et al.
Publicado: (2024)

Inducing High Energy-Latency of Large Vision-Language Models with Verbose Images
por: Gao, Kuofeng, et al.
Publicado: (2024)

MultiPriv: Benchmarking Individual-Level Privacy Reasoning in Vision-Language Models
por: Sun, Xiongtao, et al.
Publicado: (2025)

IAG: Input-aware Backdoor Attack on VLM-based Visual Grounding
por: Li, Junxian, et al.
Publicado: (2025)