Copertina

AM$^3$Safety: Towards Data Efficient Alignment of Multi-modal Multi-turn Safety for MLLMs

Fuente: arXiv

Salvato in:

Dettagli Bibliografici
Autori principali:	Zhu, Han, Chen, Jiale, Cai, Chengkun, Sun, Shengjie, Li, Haoran, Zhou, Yujin, Chan, Chi-Min, Wen, Pengcheng, Li, Lei, Han, Sirui, Guo, Yike
Natura:	Preprint
Pubblicazione:	2026
Soggetti:	Computation and Language
Accesso online:	Acceder al recurso
Tags:	Aggiungi Tag Nessun Tag, puoi essere il primo ad aggiungerne!!

Documenti analoghi

SafeMT: Multi-turn Safety for Multimodal Language Models
di: Zhu, Han, et al.
Pubblicazione: (2025)

Not Just the Destination, But the Journey: Reasoning Traces Causally Shape Generalization Behaviors
di: Wen, Pengcheng, et al.
Pubblicazione: (2026)

PKU-SafeRLHF: Towards Multi-Level Safety Alignment for LLMs with Human Preference
di: Ji, Jiaming, et al.
Pubblicazione: (2024)

Glance-or-Gaze: Incentivizing LMMs to Adaptively Focus Search via Reinforcement Learning
di: Bai, Hongbo, et al.
Pubblicazione: (2026)

MTSA: Multi-turn Safety Alignment for LLMs through Multi-round Red-teaming
di: Guo, Weiyang, et al.
Pubblicazione: (2025)

What, Whether and How? Unveiling Process Reward Models for Thinking with Images Reasoning
di: Zhou, Yujin, et al.
Pubblicazione: (2026)

ThinkPatterns-21k: A Systematic Study on the Impact of Thinking Patterns in LLMs
di: Wen, Pengcheng, et al.
Pubblicazione: (2025)

Pragma-VL: Towards a Pragmatic Arbitration of Safety and Helpfulness in MLLMs
di: Wen, Ming, et al.
Pubblicazione: (2026)

Safe RLHF-V: Safe Reinforcement Learning from Multi-modal Human Feedback
di: Ji, Jiaming, et al.
Pubblicazione: (2025)

The Side Effects of Being Smart: Safety Risks in MLLMs' Multi-Image Reasoning
di: Chen, Renmiao, et al.
Pubblicazione: (2026)

The Forgotten Shield: Safety Grafting in Parameter-Space for Medical MLLMs
di: Zhao, Jiale, et al.
Pubblicazione: (2025)

Towards Understanding Safety Alignment: A Mechanistic Perspective from Safety Neurons
di: Chen, Jianhui, et al.
Pubblicazione: (2024)

Generative RLHF-V: Learning Principles from Multi-modal Human Preference
di: Zhou, Jiayi, et al.
Pubblicazione: (2025)

State-Dependent Safety Failures in Multi-Turn Language Model Interaction
di: Li, Pengcheng, et al.
Pubblicazione: (2026)

J1: Exploring Simple Test-Time Scaling for LLM-as-a-Judge
di: Chan, Chi-Min, et al.
Pubblicazione: (2025)

HKGAI-V1: Towards Regional Sovereign Large Language Model for Hong Kong
di: Han, Sirui, et al.
Pubblicazione: (2025)

Speak Out of Turn: Safety Vulnerability of Large Language Models in Multi-turn Dialogue
di: Zhou, Zhenhong, et al.
Pubblicazione: (2024)

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

CSR-Bench: A Benchmark for Evaluating the Cross-modal Safety and Reliability of MLLMs
di: Liu, Yuxuan, et al.
Pubblicazione: (2026)

LABSHIELD: A Multimodal Benchmark for Safety-Critical Reasoning and Planning in Scientific Laboratories
di: Sun, Qianpu, et al.
Pubblicazione: (2026)

From Specific-MLLMs to Omni-MLLMs: A Survey on MLLMs Aligned with Multi-modalities
di: Jiang, Shixin, et al.
Pubblicazione: (2024)

Immune: Improving Safety Against Jailbreaks in Multi-modal LLMs via Inference-Time Alignment
di: Ghosal, Soumya Suvra, et al.
Pubblicazione: (2024)

MultiBreak: A Scalable and Diverse Multi-turn Jailbreak Benchmark for Evaluating LLM Safety
di: Song, Jialin, et al.
Pubblicazione: (2026)

Rethinking Patient Education as Multi-turn Multi-modal Interaction
di: Yao, Zonghai, et al.
Pubblicazione: (2026)

ContextLens: Modeling Imperfect Privacy and Safety Context for Legal Compliance
di: Li, Haoran, et al.
Pubblicazione: (2026)

SafeLawBench: Towards Safe Alignment of Large Language Models
di: Cao, Chuxue, et al.
Pubblicazione: (2025)

Benchmarking Multi-National Value Alignment for Large Language Models
di: Shi, Weijie, et al.
Pubblicazione: (2025)

SafeSteer: Localized On-Policy Distillation for Efficient Safety Alignment
di: Li, Hao, et al.
Pubblicazione: (2026)

CRAG-MM: Multi-modal Multi-turn Comprehensive RAG Benchmark
di: Wang, Jiaqi, et al.
Pubblicazione: (2025)

Improving the Reasoning of Multi-Image Grounding in MLLMs via Reinforcement Learning
di: Zhang, Bob, et al.
Pubblicazione: (2025)

Hard to Read, Easy to Jailbreak: How Visual Degradation Bypasses MLLM Safety Alignment
di: Song, Zhixue, et al.
Pubblicazione: (2026)

GSPR: Aligning LLM Safeguards as Generalizable Safety Policy Reasoners
di: Li, Haoran, et al.
Pubblicazione: (2025)

Reimagining Legal Fact Verification with GenAI: Toward Effective Human-AI Collaboration
di: Han, Sirui, et al.
Pubblicazione: (2026)

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

Multi-Prompt Alignment for Multi-Source Unsupervised Domain Adaptation
di: Chen, Haoran, et al.
Pubblicazione: (2022)

Think in Safety: Unveiling and Mitigating Safety Alignment Collapse in Multimodal Large Reasoning Model
di: Lou, Xinyue, et al.
Pubblicazione: (2025)

Reasoned Safety Alignment: Ensuring Jailbreak Defense via Answer-Then-Check
di: Cao, Chentao, et al.
Pubblicazione: (2025)

Thinking by Doing: Building Efficient World Model Reasoning in LLMs via Multi-turn Interaction
di: Shu, Bao, et al.
Pubblicazione: (2025)

Global Tropical Cyclone Intensity Forecasting with Multi-modal Multi-scale Causal Autoregressive Model
di: Wang, Xinyu, et al.
Pubblicazione: (2024)

Do We Really Need Curated Malicious Data for Safety Alignment in Multi-modal Large Language Models?
di: Wang, Yanbo, et al.
Pubblicazione: (2025)