The paper introduces SEAL, a training-time, parameter‑efficient defense that attaches a plug‑and‑play adapter to the shared expert component of Mixture‑of‑Experts models, and SEAL++, which adds an orthogonal constraint to preserve existing safety subspaces. By leveraging the always‑activated shared expert, SEAL mitigates the structural vulnerability of sparse routing to adversarial manipulation, reducing attack success rates by up to 60% with minimal impact on model capability. The approach is evaluated across six attack scenarios involving harmful prompting, jailbreaks, malicious fine‑tuning, and neuron pruning.
By Qingyu Meng, Yiwei Zha, Jiahuan Pei, Koen Hindriks, Herbert Bos, Min Chen
arXiv:2609.00051v1 Announce Type: cross
Abstract: Despite extensive alignment efforts, Large Language Models (LLMs) remain vulnerable to generating unsafe content under adversarial prompting, yet the...
By Kuan-Lin Chu, Chung-En Sun, Tsui-Wei Weng
arXiv:2607. 27386v1 Announce Type: cross Abstract: Diffusion Language Models (DLMs) offer a compelling alternative to autoregressive (AR) generation by enabling bidirectional context and iterative refinement.
By Saurabh Yadav, Badri Narayana Patro, Vijay Srinivas Agneeswaran
The paper addresses the problem of large language models (LLMs) over-refusing to comply with benign safety-related instructions. It identifies that a small number of hypersensitive safety heads in transformer attention misfire on hard-safe prompts, causing abnormal attention entanglement and high-entropy routing conflicts that lead to refusals. The authors propose Semantic Routing Calibration (SRC), a lightweight, training‑free inference framework that dynamically suppresses these problematic heads and fuses logits to restore trustworthy reasoning while preserving intrinsic safety performance.
By Zixuan Wang, Bingjie Zhang, He Zhao, Dandan Guo
The paper addresses the problem of over‑refusal in safety‑aligned large language models, where benign instructions are incorrectly rejected. It identifies that a small set of hypersensitive safety heads in transformer attention misfire on hard‑safe prompts, causing abnormal attention entanglement and high‑entropy routing conflicts that block necessary attention to target entities. To mitigate this, the authors propose Semantic Routing Calibration (SRC), a lightweight, training‑free inference framework that dynamically suppresses these hypersensitive heads and fuses logits from dual branches to restore trustworthy reasoning while preserving intrinsic safety performance.
By Zixuan Wang, Bingjie Zhang, He Zhao, Dandan Guo
arXiv:2607. 08883v1 Announce Type: new Abstract: Behavioral alignment in large language models often masks fragile internal safety representations.
By Ege \c{C}akar, Hannah Guan, Kayden Kehe
The paper introduces Dynamic Semantic Steering (DSS), a training‑free, inference‑time defense for robust concept erasure in text‑to‑image diffusion models. DSS models local semantic neighborhoods geometrically, automatically identifies benign semantic anchors, and applies context‑aware, constrained feature correction using cross‑attention signals. Experiments show DSS achieves an average erasure rate of 91.0%, outperforming prior defenses while reducing semantic drift and preserving generation fidelity.
By Qinghui Gong, Zhengchun Zhou, Hua Meng, Yihuai Liang, Yuxuan Zhang
UpSafe°C is a unified framework that enhances large language model safety by upcycling safety‑critical layers into a sparse Mixture‑of‑Experts structure, with a router acting as a soft guardrail. It employs a two‑stage supervised fine‑tuning strategy to improve safety discrimination while preserving general capabilities, and introduces a safety temperature mechanism for dynamic inference‑time control of the safety‑utility trade‑off. Experiments across multiple benchmarks and model scales show robust safety improvements against harmful and jailbreak inputs while maintaining competitive general performance.
By Yuhao Sun, Zhuoer Xu, Shiwen Cui, Kun Yang, Lingyun Yu, Yongdong Zhang, Hongtao Xie
arXiv:2610.07774v1 Announce Type: new
Abstract: Vision-language models (VLMs) face compositional safety risks where harmful intent emerges from the interaction between visual and textual inputs. As m...
By Ziyuan Yang, Wenxuan Ding, Shangbin Feng, Yulia Tsvetkov
arXiv:2607. 01859v1 Announce Type: new Abstract: Safety training for large language models (LLMs) is conducted predominantly in English, leaving uncertain how well safety mechanisms generalize to low-resource languages and mixed-language code-switching.
By Joshua Adrian Cahyono
arXiv:2607. 07903v1 Announce Type: cross Abstract: Large language models (LLMs) exhibit remarkable capabilities but remain highly vulnerable to adversarial prompts and jailbreak attacks.
By Anupam Wagle, Ifrat Ikhtear Uddin, Chaowei Zhang, Longwei Wang
arXiv:2604. 23130v2 Announce Type: replace-cross Abstract: Jailbreak attacks expose a persistent failure mode in safety-aligned LLMs: models can be pushed into harmful behavior, but the internal representations enabling this shift remain poorly localized.
By Nilanjana Das, Mathew Dawit, Aman Chadha, Manas Gaur