paper-with-me

Papers

PRISM: Robust VLM Alignment with Principled Reasoning for Integrated Safety in Multimodality

2025-08-26 · Nanxi Li, Zhengyue Zhao, G. Edward Suh, Marco Pavone, Chaowei Xiao arxiv

Safeguarding vision-language models (VLMs) is a critical challenge, as existing methods often suffer from over-defense, which harms utility, or rely on shallow alignment, failing to detect complex threats that require deep reasoning. To this end, we introduc PRISM (Principled Reasoning for Integrated Safety in Multimodality), a System 2-like framework that aligns VLMs through a structured four-stage reasoning process explicitly designed to handle three distinct categories of multimodal safety violations. Our framework consists of two key components: a structured reasoning pipeline that analyzes each violation category in dedicated stages, and PRISM-DPO, generated via Monte Carlo Tree Search (MCTS) to refine reasoning quality through Direct Preference Optimization. Comprehensive evaluations show that PRISM substantially reduces attack success rates on JailbreakV-28K and VLBreak, improves robustness against adaptive attacks, and generalizes to out-of-distribution multi-image threats, while better preserving model utility on benign multimodal benchmarks. Our code, data, and model weights available at https://github.com/SaFoLab-WISC/PRISM.

📄 PDF Abstract BibTeX arXiv:2508.18649

Code (0)

등록된 구현이 없습니다.

Similar Papers 제목 키워드 기반

Alignment-Weighted DPO: A principled reasoning approach to improve safety alignment

2026-02-24 · Mengxuan Hu, Vivek V. Datla, Anoop Kumar, Zihan Guan 외 arxiv

Recent advances in alignment techniques such as Supervised Fine-Tuning (SFT), Reinforcement Learning from Human Feedback (RLHF), and Direct Preference Optimization (DPO) have improved the safety of large language models …

Reinforcement Learning

Expert Personas Improve LLM Alignment but Damage Accuracy: Bootstrapping Intent-Based Persona Routing with PRISM

2026-03-19 · Zizhao Hu, Mohammad Rostami, Jesse Thomason arxiv

Persona prompting can steer LLM generation towards a domain-specific tone and pattern. This behavior enables use cases in multi-agent systems where diverse interactions are crucial and human-centered tasks require high-l…

Large Reasoning Models Learn Better Alignment from Flawed Thinking

2025-10-01 · ShengYun Peng, Eric Smith, Ivan Evtimov, Song Jiang 외 arxiv

Large reasoning models (LRMs) "think" by generating structured chain-of-thought (CoT) before producing a final answer, yet they still lack the ability to reason critically about safety alignment and are easily biased whe…

Reinforcement Learning

PRISM: Prosody-Integrated Multi-Agent Reasoning Framework for Empathetic Spoken Dialogue

2026-06-11 · Wen Zhang, Xiaocui Yang, Zhuoyue Gao, Shi Feng 외 arxiv

Empathetic spoken dialogue systems require not only semantically appropriate responses but also emotionally aligned prosodic expression. However, cascade pipelines often discard acoustic cues during speech-to-text conver…

Dialogue GenerationResponse GenerationSpeech Synthesis

Beyond SFT-to-RL: Pre-alignment via Black-Box On-Policy Distillation for Multimodal RL

2026-04-30 · Sudong Wang, Weiquan Huang, Xiaomin Yu, Zuhao Yang 외 arxiv

The standard post-training recipe for large multimodal models (LMMs) applies supervised fine-tuning (SFT) on curated demonstrations followed by reinforcement learning with verifiable rewards (RLVR). However, SFT introduc…

Reinforcement LearningMultimodal ReasoningVisual Grounding