Cover Image

The Side Effects of Being Smart: Safety Risks in MLLMs' Multi-Image Reasoning

Fuente: arXiv

Saved in:

Bibliographic Details
Main Authors:	Chen, Renmiao, Lu, Yida, Cui, Shiyao, Ouyang, Xuan, Huang, Victor Shea-Jay, Zhang, Shumin, Pan, Chengwei, Qiu, Han, Huang, Minlie
Format:	Preprint
Published:	2026
Subjects:	Computer Vision and Pattern Recognition Computation and Language
Online Access:	Acceder al recurso
Tags:	Add Tag No Tags, Be the first to tag this record!

Similar Items

ShieldVLM: Safeguarding the Multimodal Implicit Toxicity via Deliberative Reasoning with LVLMs
by: Cui, Shiyao, et al.
Published: (2025)

JPS: Jailbreak Multimodal Large Language Models with Collaborative Visual Perturbation and Textual Steering
by: Chen, Renmiao, et al.
Published: (2025)

BlackDAN: A Black-Box Multi-Objective Approach for Effective and Contextual Jailbreaking of Large Language Models
by: Wang, Xinyuan, et al.
Published: (2024)

Agent-SafetyBench: Evaluating the Safety of LLM Agents
by: Zhang, Zhexin, et al.
Published: (2024)

AISafetyLab: A Comprehensive Framework for AI Safety Evaluation and Improvement
by: Zhang, Zhexin, et al.
Published: (2025)

How Should We Enhance the Safety of Large Reasoning Models: An Empirical Study
by: Zhang, Zhexin, et al.
Published: (2025)

SMSP: A Plug-and-Play Strategy of Multi-Scale Perception for MLLMs to Perceive Visual Illusions
by: Tu, Jinzhe, et al.
Published: (2026)

The Missing Half: Unveiling Training-time Implicit Safety Risks Beyond Deployment
by: Zhang, Zhexin, et al.
Published: (2026)

LongSafety: Evaluating Long-Context Safety of Large Language Models
by: Lu, Yida, et al.
Published: (2025)

Survive at All Costs: Exploring LLM's Risky Behaviors under Survival Pressure
by: Lu, Yida, et al.
Published: (2026)

Exploring Multimodal Challenges in Toxic Chinese Detection: Taxonomy, Benchmark, and Findings
by: Yang, Shujian, et al.
Published: (2025)

From Theft to Bomb-Making: The Ripple Effect of Unlearning in Defending Against Jailbreak Attacks
by: Zhang, Zhexin, et al.
Published: (2024)

The Superalignment of Superhuman Intelligence with Large Language Models
by: Huang, Minlie, et al.
Published: (2024)

When Smiley Turns Hostile: Interpreting How Emojis Trigger LLMs' Toxicity
by: Cui, Shiyao, et al.
Published: (2025)

LASA: Language-Agnostic Semantic Alignment at the Semantic Bottleneck for LLM Safety
by: Yang, Junxiao, et al.
Published: (2026)

Be Careful When Fine-tuning On Open-Source LLMs: Your Fine-tuning Data Could Be Secretly Stolen!
by: Zhang, Zhexin, et al.
Published: (2025)

Understanding the Dark Side of LLMs' Intrinsic Self-Correction
by: Zhang, Qingjie, et al.
Published: (2024)

Towards Exception Safety Code Generation with Intermediate Representation Agents Framework
by: Zhang, Xuanming, et al.
Published: (2024)

Stop Before You Fail: Operational Capability Boundaries for Mitigating Unproductive Reasoning in Large Reasoning Models
by: Zhang, Qingjie, et al.
Published: (2025)

ShieldLM: Empowering LLMs as Aligned, Customizable and Explainable Safety Detectors
by: Zhang, Zhexin, et al.
Published: (2024)

MM-THEBench: Do Reasoning MLLMs Think Reasonably?
by: Huang, Zhidian, et al.
Published: (2026)

The Instinctive Bias: Spurious Images lead to Illusion in MLLMs
by: Han, Tianyang, et al.
Published: (2024)

SafetyBench: Evaluating the Safety of Large Language Models
by: Zhang, Zhexin, et al.
Published: (2023)

Improving Text-to-Image Generation with Input-Side Inference-Time Scaling
by: Chen, Ruibo, et al.
Published: (2025)

BARREL: Boundary-Aware Reasoning for Factual and Reliable LRMs
by: Yang, Junxiao, et al.
Published: (2025)

Towards Faithful Reasoning in Comics for Small MLLMs
by: Feng, Chengcheng, et al.
Published: (2026)

Unlocking Reasoning Potential in Large Langauge Models by Scaling Code-form Planning
by: Wen, Jiaxin, et al.
Published: (2024)

SpaceR: Reinforcing MLLMs in Video Spatial Reasoning
by: Ouyang, Kun, et al.
Published: (2025)

Aesthetic Image Captioning with Saliency Enhanced MLLMs
by: Tao, Yilin, et al.
Published: (2025)

EgoMind: Activating Spatial Cognition through Linguistic Reasoning in MLLMs
by: Chen, Zhenghao, et al.
Published: (2026)

Can MLLMs Read the Room? A Multimodal Benchmark for Assessing Deception in Multi-Party Social Interactions
by: Kang, Caixin, et al.
Published: (2025)

Can MLLMs Read the Room? A Multimodal Benchmark for Verifying Truthfulness in Multi-Party Social Interactions
by: Kang, Caixin, et al.
Published: (2025)

PatientHub: A Unified Framework for Patient Simulation
by: Sabour, Sahand, et al.
Published: (2026)

Distill Visual Chart Reasoning Ability from LLMs to MLLMs
by: He, Wei, et al.
Published: (2024)

MMReason: An Open-Ended Multi-Modal Multi-Step Reasoning Benchmark for MLLMs Toward AGI
by: Yao, Huanjin, et al.
Published: (2025)

Latent Sketchpad: Sketching Visual Thoughts to Elicit Multimodal Reasoning in MLLMs
by: Zhang, Huanyu, et al.
Published: (2025)

Can MLLMs Understand the Deep Implication Behind Chinese Images?
by: Zhang, Chenhao, et al.
Published: (2024)

Seeker: Towards Exception Safety Code Generation with Intermediate Language Agents Framework
by: Zhang, Xuanming, et al.
Published: (2024)

AM$^3$Safety: Towards Data Efficient Alignment of Multi-modal Multi-turn Safety for MLLMs
by: Zhu, Han, et al.
Published: (2026)

Think Socially via Cognitive Reasoning
by: Zhou, Jinfeng, et al.
Published: (2025)