arXiv AI June 2, 2026 By Yitong Sun, Yao Huang, Teng Li, Ranjie Duan, Yichi Zhang, Xingjun Ma, Hui Xue, Xingxing Wei

MESA: Improving MoE Safety Alignment via Decentralized Expertise

arXiv:2606. 00651v1 Announce Type: cross Abstract: Mixture-of-Experts (MoE) architectures scale Large Language Models (LLMs) efficiently, enabling greater capacity with reduced computational cost by dynamically routing inputs to relevant experts, yet introduce a critical vulnerability: Safety Sparsity, where safety capabilities concentrate in few experts, making them susceptible to adversarial bypassing.

Summary generated by The Flow from the publisher's feed. The full article lives at arXiv AI.

arXiv Machine Learning

Jul 30

On-Policy Distillation for LLM Safety: A Routing Approach to Template-Robust Realignment

arXiv:2607. 27081v1 Announce Type: cross Abstract: Fine-tuning is the dominant paradigm for specializing large language models (LLMs), yet it exposes a critical vulnerability: malicious data providers can embed harmful behaviors into downstream corpora, creating models that retain professional skills while violating human values on demand.

By Yongjian Guo, Wanlun Ma, Lingyu Shen, Xi Xiao, Sheng Wen

llms fine-tuning efficiency benchmarks safety

arXiv Machine Learning

5d ago

HiRoute: Hierarchical Routed Prompt Tuning for Safety Alignment of Large Language Models

arXiv:2608. 12821v1 Announce Type: new Abstract: Large language models (LLMs) remain vulnerable to harmful requests and jailbreak attacks.

By Fangzhou Chen, Shiji Zhao, Mengyang Wang, Qihui Zhu, Ranjie Duan, Maoxun Yuan, Xingxing Wei

llms rag benchmarks safety

arXiv Machine Learning

Jun 2

Safety Game: Inference-Time Alignment of Black-Box LLMs via Constrained Optimization

arXiv:2510. 09330v3 Announce Type: replace Abstract: Ensuring that large language models (LLMs) comply with safety requirements is a central challenge in AI deployment.

By Tuan Nguyen, Long Tran-Thanh

llms agents reinforcement-learning fine-tuning safety

arXiv Machine Learning

Jun 2

Dialectics of Alignment: Harnessing Unsafe Knowledge for Dynamic Safety Routing

arXiv:2606. 00686v1 Announce Type: new Abstract: The prevailing paradigm in large language model (LLM) alignment operates via erasure, filtering unsafe data or training models to strictly refuse harmful prompts.

By Maryam Hashemzadeh, Jerry Huang, Minseon Kim, Marc-Alexandre C\^ot\'e, Sarath Chandar

llms fine-tuning benchmarks safety

arXiv AI

Aug 5

Moving the Safety Barrier: Dynamic Routing Adaptive Alignment Against White-Box Attacks

arXiv:2608. 02674v1 Announce Type: cross Abstract: With the widespread deployment of large foundation models (LFMs) in open environments, safety threats are shifting from black-box jailbreaks toward white-box attacks that directly identify and disrupt internal safety neurons or routes.

By Shangze Li, Chuancheng Shi, Simiao Xie, Lingzhi He, Cheng Ji, Zifeng Cheng, Fei Shen, Chao Wu, Tat-Seng Chua

safety

arXiv AI

Jul 21

TRACE: Trajectory-Based Safety Patch Learning for LLM Post-Training Realignment

arXiv:2607. 16242v1 Announce Type: cross Abstract: Fine-Tuning-as-a-Service (FTaaS) platforms let users train large language models (LLMs) on customized tasks, but this pipeline could erode models' safety alignment.

By Changyue Li, Jiaming He, Youliang Yuan, Jialin Wu, Boxi Yu, Zhicong Huang, Pinjia He

llms fine-tuning benchmarks safety

Related stories

On-Policy Distillation for LLM Safety: A Routing Approach to Template-Robust Realignment

HiRoute: Hierarchical Routed Prompt Tuning for Safety Alignment of Large Language Models

Safety Game: Inference-Time Alignment of Black-Box LLMs via Constrained Optimization

Dialectics of Alignment: Harnessing Unsafe Knowledge for Dynamic Safety Routing

Moving the Safety Barrier: Dynamic Routing Adaptive Alignment Against White-Box Attacks

TRACE: Trajectory-Based Safety Patch Learning for LLM Post-Training Realignment