arXiv subjects

Fenghua Weng

Publications and source records attributed to Fenghua Weng.

6 recordsLinked to original sources

arXiv2025

Think-Reflect-Revise: A Policy-Guided Reflective Framework for Safety Alignment in Large Vision Language Models

As multimodal reasoning improves the overall capabilities of Large Vision Language Models (LVLMs), recent studies have begun to explore safety-oriented reasoning, aiming to enhance safety awareness by analyzing potential safety risks during the reasoning process before generating the final response. Although such approaches improve safety awareness and interpretability, this single-pass think-then-answer paradigm remains vulnerable to contextual or visual jailbreak attacks. This reveals a critical flaw: single-pass reasoning may overlook explicit harmful content in its own output. Our key insight is to exploit this wasted signal through reflection, which can effectively leverage the malicious content revealed in the first-pass reasoning to enable genuine self-correction and prevent unsafe generations. Motivated by this, we propose Think-Reflect-Revise (TRR), a three-stage training framework designed to enhance the safety alignment of LVLMs through policy-guided self-reflection. We first build a Reflective Safety Reasoning (ReSafe) dataset with 5,000 examples that follow a think-reflect-revise process. We then fine-tune the target model using the ReSafe dataset to initialize reflective behavior, and finally reinforce policy-guided reflection through reinforcement learning. Experimental results show that TRR substantially improves the safety performance of LVLMs across both safety-awareness benchmarks and jailbreak attack evaluations, increasing the overall safe response rate from 42.8% to 87.7% on Qwen2.5-VL-7B, while preserving stable performance on general benchmarks such as MMMU and MMStar. The project page is available at https://think-reflect-revise.github.io/.

cs.CV↗

arXiv2025

COSMO-RL: Towards Trustworthy LMRMs via Joint Safety and Stability

Large Multimodal Reasoning Models (LMRMs) are moving into real applications, where they must be both useful and safe. Safety is especially challenging in multimodal settings: images and text can be combined to bypass guardrails, and single objective training can cause policy drift that yields over-refusal on benign inputs or unsafe compliance on risky ones. We present COSMO-RL, a mixed reinforcement learning framework that trains reasoning oriented LMRMs under multimodal, multitask, and multiobjective signals, and we release the resulting model, COSMO-R1. Our approach aims to let safety and capability grow together in one stable pipeline rather than competing during alignment. In experiments, COSMO-R1 improves safety while maintaining-and often improving multimodal reasoning and instruction following, shows stronger robustness to multimodal jailbreaks, and reduces unnecessary refusals. The framework also transfers across backbones with consistent gains. Ablations support the design choices, indicating a simple path to advancing safety and general capability together in LMRMs.

cs.AI↗

arXiv2025

SafeWork-R1: Coevolving Safety and Intelligence under the AI-45$^{\circ}$ Law

We introduce SafeWork-R1, a cutting-edge multimodal reasoning model that demonstrates the coevolution of capabilities and safety. It is developed by our proposed SafeLadder framework, which incorporates large-scale, progressive, safety-oriented reinforcement learning post-training, supported by a suite of multi-principled verifiers. Unlike previous alignment methods such as RLHF that simply learn human preferences, SafeLadder enables SafeWork-R1 to develop intrinsic safety reasoning and self-reflection abilities, giving rise to safety `aha' moments. Notably, SafeWork-R1 achieves an average improvement of $46.54\%$ over its base model Qwen2.5-VL-72B on safety-related benchmarks without compromising general capabilities, and delivers state-of-the-art safety performance compared to leading proprietary models such as GPT-4.1 and Claude Opus 4. To further bolster its reliability, we implement two distinct inference-time intervention methods and a deliberative search mechanism, enforcing step-level verification. Finally, we further develop SafeWork-R1-InternVL3-78B, SafeWork-R1-DeepSeek-70B, and SafeWork-R1-Qwen2.5VL-7B. All resulting models demonstrate that safety and capability can co-evolve synergistically, highlighting the generalizability of our framework in building robust, reliable, and trustworthy general-purpose AI.

cs.AI↗

arXiv2025

DELMAN: Dynamic Defense Against Large Language Model Jailbreaking with Model Editing

Large Language Models (LLMs) are widely applied in decision making, but their deployment is threatened by jailbreak attacks, where adversarial users manipulate model behavior to bypass safety measures. Existing defense mechanisms, such as safety fine-tuning and model editing, either require extensive parameter modifications or lack precision, leading to performance degradation on general tasks, which is unsuitable to post-deployment safety alignment. To address these challenges, we propose DELMAN (Dynamic Editing for LLMs JAilbreak DefeNse), a novel approach leveraging direct model editing for precise, dynamic protection against jailbreak attacks. DELMAN directly updates a minimal set of relevant parameters to neutralize harmful behaviors while preserving the model's utility. To avoid triggering a safe response in benign context, we incorporate KL-divergence regularization to ensure the updated model remains consistent with the original model when processing benign queries. Experimental results demonstrate that DELMAN outperforms baseline methods in mitigating jailbreak attacks while preserving the model's utility, and adapts seamlessly to new attack instances, providing a practical and efficient solution for post-deployment model protection.

cs.CR↗

arXiv2025

Adversary-Aware DPO: Enhancing Safety Alignment in Vision Language Models via Adversarial Training

Safety alignment is critical in pre-training large language models (LLMs) to generate responses aligned with human values and refuse harmful queries. Unlike LLM, the current safety alignment of VLMs is often achieved with post-hoc safety fine-tuning. However, these methods are less effective to white-box attacks. To address this, we propose $\textit{Adversary-aware DPO (ADPO)}$, a novel training framework that explicitly considers adversarial. $\textit{Adversary-aware DPO (ADPO)}$ integrates adversarial training into DPO to enhance the safety alignment of VLMs under worst-case adversarial perturbations. $\textit{ADPO}$ introduces two key components: (1) an adversarial-trained reference model that generates human-preferred responses under worst-case perturbations, and (2) an adversarial-aware DPO loss that generates winner-loser pairs accounting for adversarial distortions. By combining these innovations, $\textit{ADPO}$ ensures that VLMs remain robust and reliable even in the presence of sophisticated jailbreak attacks. Extensive experiments demonstrate that $\textit{ADPO}$ outperforms baselines in the safety alignment and general utility of VLMs.

cs.CR↗

arXiv2024

$\textit{MMJ-Bench}$: A Comprehensive Study on Jailbreak Attacks and Defenses for Multimodal Large Language Models

As deep learning advances, Large Language Models (LLMs) and their multimodal counterparts, Multimodal Large Language Models (MLLMs), have shown exceptional performance in many real-world tasks. However, MLLMs face significant security challenges, such as jailbreak attacks, where attackers attempt to bypass the model's safety alignment to elicit harmful responses. The threat of jailbreak attacks on MLLMs arises from both the inherent vulnerabilities of LLMs and the multiple information channels that MLLMs process. While various attacks and defenses have been proposed, there is a notable gap in unified and comprehensive evaluations, as each method is evaluated on different dataset and metrics, making it impossible to compare the effectiveness of each method. To address this gap, we introduce \textit{MMJ-Bench}, a unified pipeline for evaluating jailbreak attacks and defense techniques for MLLMs. Through extensive experiments, we assess the effectiveness of various attack methods against SoTA MLLMs and evaluate the impact of defense mechanisms on both defense effectiveness and model utility for normal tasks. Our comprehensive evaluation contribute to the field by offering a unified and systematic evaluation framework and the first public-available benchmark for MLLM jailbreak research. We also demonstrate several insightful findings that highlights directions for future studies.

cs.CR↗