VisuoAlign: Safety Alignment of LVLMs with Multimodal Tree Search
On the map: LLM reasoning → Reasoning for agents, planning and code
also in 3 more
Abstract (arXiv)
Large Vision-Language Models (LVLMs) have achieved remarkable progress in multimodal perception and generation, yet their safety alignment remains a critical challenge.Existing defenses and vulnerable to multimodal jailbreaks, as visual inputs introduce new attack surfaces, reasoning chains lack safety supervision, and alignment often degrades under modality fusion.To overcome these limitation, we propose VisuoAlign, a framework for multi-modal safety alignment via prompt-guided tree search.VisuoAlign embeds safety constrains into the reasoning process through visual-textual interactive prompts, employs Monte Carlo Tree Search(MCTS) to systematically construct diverse safety-critical prompt trajectories, and introduces prompt-based scaling to ensure real-time risk detection and compliant responses.Extensive experiments demonstrate that VisuoAlign proactively exposes risks, enables comprehensive dataset generation, and significantly improves the robustness of LVLMs against complex cross-modal threats.
Key facts. No citing papers in the map yet (data through 7 Oct 2026). Placed in Reasoning for agents, planning and code, one of 17 directions in LLM reasoning. Builds on 38 of its 45 references that are in the map. First posted on arXiv 10 Oct 2025.
Builds on
38 of 45 references.
Closest ancestors · 2
Same direction, newest first
- 2025Safety in Large Reasoning Models: A Survey24 Apr
- 2025SafeMLRM: Demystifying Safety in Multi-modal Large Reasoning Models9 Apr
Other references · 36
- 2022Training language models to follow instructions with human feedbackcited by 15,712
- 2023Direct Preference Optimization: Your Language Model is Secretly a Reward Modelcited by 7,916
- 2023Visual Instruction Tuningcited by 7,792
- 2017Towards Deep Learning Models Resistant to Adversarial Attackscited by 6,326
- 2022LAION-5B: An open large-scale dataset for training next generation image-text modelscited by 4,131
- 2023Qwen Technical Reportcited by 3,273
- 2023Tree of Thoughts: Deliberate Problem Solving with Large Language Modelscited by 3,155
- 2023InstructBLIP: Towards General-purpose Vision-Language Models with Instruction Tuningcited by 2,855
- 2023MiniGPT-4: Enhancing Vision-Language Understanding with Advanced Large Language Modelscited by 2,689
- 2024LLaVA-NeXT-Interleave: Tackling Multi-image, Video, and 3D in Large Multimodal Modelscited by 571
- 2023Visual Adversarial Examples Jailbreak Aligned Large Language Modelscited by 307
- 2023MM-SafetyBench: A Benchmark for Safety Evaluation of Multimodal Large Language Modelscited by 255
24 more
- 2024SPA-VL: A Comprehensive Safety Preference Alignment Dataset for Vision Language Modelcited by 82
- 2024On the Role of Attention Heads in Large Language Model Safetycited by 70
- 2024Heuristic-Induced Multimodal Risk Distribution Jailbreak Attack for Multimodal Large Language Modelscited by 29
- 2024Unraveling and Mitigating Safety Alignment Degradation of Vision-Language Modelscited by 23
- 2024Reinforcement Learning from Multi-role Debates as Feedback for Bias Mitigation in LLMscited by 22
- 2025Inverse Reinforcement Learning with Dynamic Reward Scaling for LLM Alignmentcited by 20
- 2023Making Harmful Behaviors Unlearnable for Large Language Modelscited by 19
- 2025Oyster-I: Beyond Refusal -- Constructive Safety Alignment for Responsible Language Modelscited by 19
- 2024Randomized Exploration in Cooperative Multi-Agent Reinforcement Learningcited by 17
- 2024PBI-Attack: Prior-Guided Bimodal Interactive Black-Box Jailbreak Attack for Toxicity Maximizationcited by 17
- 2024Upper and Lower Bounds for Distributionally Robust Off-Dynamics Reinforcement Learningcited by 15
- 2024AGR: Age Group fairness Reward for Bias Mitigation in LLMscited by 12
- 2025FAEDKV: Infinite-Window Fourier Transform for Unbiased KV Cache Compressioncited by 11
- 2025Strata-Sword: A Hierarchical Safety Evaluation towards LLMs based on Reasoning Complexity of Jailbreak Instructionscited by 11
- 2024Gibberish is All You Need for Membership Inference Detection in Contrastive Language-Audio Pretrainingcited by 10
- 2025MedGUIDE: Benchmarking Clinical Decision-Making in Large Language Modelscited by 10
- 2025When Truthful Representations Flip Under Deceptive Instructions?cited by 9
- 2024TUNI: A Textual Unimodal Detector for Identity Inference in CLIP Modelscited by 8
- 2025Quantized but Deceptive? A Multi-Dimensional Truthfulness Evaluation of Quantized LLMscited by 8
- 2025Cross-Modal Obfuscation for Jailbreak Attacks on Large Vision-Language Modelscited by 5
- 2025A Survey On Secure Machine Learningcited by 3
- 2025L-CLIPScore: a Lightweight Embedding-based Captioning Metric for Evaluating and Trainingcited by 2
- 2025Pruning Weights but Not Truth: Safeguarding Truthfulness While Pruning LLMscited by 2
- 2025Adversarial Attacks on VQA-NLE: Exposing and Alleviating Inconsistencies in Visual Question Answering Explanationscited by 1
Built on by
Not cited yet.
Similar, not cited
Close in wording, no citation either way
- 2026GuardAlign: Test-time Safety Alignment in Multimodal Large Language Models
- 2025MSR-Align: Policy-Grounded Multimodal Alignment for Safety-Aware Reasoning in Vision-Language Models
- 2024ETA: Evaluating Then Aligning Safety of Vision Language Models at Inference Time
- 2025PRISM: Robust VLM Alignment with Principled Reasoning for Integrated Safety in Multimodality
- 2024Cross-Modal Safety Mechanism Transfer in Large Vision-Language Models
- 2025Security Tensors as a Cross-Modal Bridge: Extending Text-Aligned Safety to Vision in LVLM
- 2025DAVSP: Safety Alignment for Large Vision-Language Models via Deep Aligned Visual Safety Prompt
- 2025VLMGuard-R1: Proactive Safety Alignment for VLMs via Reasoning-Driven Prompt Optimization