FusionAudit: Pathway-Conditioned Robustness Auditing for Native Multimodal Models

arXiv:2610.21667 · cs.CV · Submitted 2026-10-15 · Read on arXiv

cs.CV

Submitted: 2026-10-15

Updated: 2026-10-15

Comments: NeurIPS 2026

Project page: https://jiamingzhang.netlify.app/fusionaudit

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Terminology

• Robustness — 1× in this paper · explained in 5 episodes

Sources

• Abusing Images and Sounds for Indirect Instruction Injection in Multi-Modal LLMs
• Qwen3-VL Technical Report
• Exploring Typographic Visual Prompts Injection Threats in Cross-Modality Generation Models
• ANOLE: An Open, Autoregressive, Native Large Multimodal Models for Interleaved Image-Text Generation
• MME: A Comprehensive Evaluation Benchmark for Multimodal Large Language Models
• DiffuGuard: How Intrinsic Safety is Lost and Found in Diffusion Large Language Models
• DeepSeek-VL: Towards Real-World Vision-Language Understanding
• VEAttack: Downstream-agnostic Vision Encoder Attack against Large Vision Language Models
• Robust CLIP: Unsupervised Adversarial Fine-Tuning of Vision Embeddings for Robust Large Vision-Language Models
• Jailbreak in pieces: Compositional Adversarial Attacks on Multi-Modal Language Models
• Chameleon: Mixed-Modal Early-Fusion Foundation Models
• Omni-Safety under Cross-Modality Conflict: Vulnerabilities, Dynamics Mechanisms and Efficient Alignment
• Double Visual Defense: Adversarial Pre-training and Instruction Tuning for Improving Vision-Language Model Robustness
• The Dawn of LMMs: Preliminary Explorations with GPT-4V(ision)
• Scaling Autoregressive Multi-Modal Models: Pretraining and Instruction Tuning
• CAVALRY-V: A Large-Scale Generator Framework for Adversarial Attacks on Video MLLMs

Related papers

• Loss Knows Best: Detecting Annotation Errors in Videos via Loss Trajectories
• AnchorWeave: World-Consistent Video Generation with Retrieved Local Spatial Memories
• Benchmarking the Robustness of Foundation Models for Mammography under Domain Shift
• MambaX-Net: Dual-Input Mamba-Enhanced Cross-Attention Network for Longitudinal MRI Segmentation
• TeleOCR: Navigating Document Parsing Across Digital and Camera-Captured Documents
• A Survey on Efficient Vision-Language-Action Models