We propose patching for large language models (LLMs) like software versions, a lightweight and modular approach for addressing safety vulnerabilities. While vendors release improved LLM versions, major releases are costly, infrequent, and difficult to tailor to customer needs, leaving released models with known safety gaps. Unlike full-model fine-tuning or major version updates, our method enables rapid remediation by prepending a compact, learnable prefix to an existing model. This "patch" introduces only 0.003% additional parameters, yet reliably steers model behavior toward that of a safer reference model. Across three critical domains (toxicity mitigation, bias reduction, and harmfulness refusal) policy patches achieve safety improvements comparable to next-generation safety-aligned models while preserving fluency. Our results demonstrate that LLMs can be "patched" much like software, offering vendors and practitioners a practical mechanism for distributing scalable, efficient, and composable safety updates between major model releases.
Q-resafe: Assessing Safety Risks and Quantization-aware Safety Patching for Quantized Large Language Models
The State Key Laboratory of Blockchain and Data Security, Zhejiang University（区块链与数据安全国家重点实验室，浙江大学）；Sun Yat-sen University（中山大学）；Hangzhou High-Tech Zone (Binjiang) Institute of Blockchain and Data Security（杭州高新技术区（滨江）区块链与数据安全研究院）
Towards Understanding Safety Alignment: A Mechanistic Perspective from Safety Neurons
Department of Computer Science and Technology, BNRist（计算机科学与技术系，BNRist）；Shenzhen International Graduate School（深圳国际研究生院）；KIRC, Institute for Artificial Intelligence, Tsinghua University（人工智能研究院，清华大学）
Light Alignment Improves LLM Safety via Model Self-Reflection with a Single Neuron
Beijing Institute of AI Safety and Governance（北京人工智能安全与治理研究院）；Beijing Key Laboratory of Safe AI and Superalignment（北京安全人工智能与超对齐重点实验室）；Ant Group Co., Ltd.（蚂蚁集团有限公司）；BrainCog Lab., CASIA（CASIA脑认知实验室）；Zhongguancun Academy（中关村学院）
CLEAR: Continuous Latent Adapter Routing for Utility-Preserving LLM Safety Alignment
Stanford University（斯坦福大学）；University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校）；Siebel School of Computing and Data Science, University of Illinois at Urbana-Champaign（伊利诺伊大学厄巴纳-香槟分校西贝尔计算与数据科学学院）；Computer Science, Stanford University（斯坦福大学计算机科学系）
ReFrame: Evidence-Guided Test-Time Safety Alignment in Multimodal Large Language Models
College of Computer Science and Technology, National University of Defense Technology（国防科技大学计算机学院）；State Key Laboratory of Complex & Critical Software Environment（复杂与关键软件环境国家重点实验室）；School of Computer Science, Wuhan University（武汉大学计算机学院）
Aligned but Fragile: Enhancing LLM Safety Robustness via Zeroth-Order Optimization
The State Key Laboratory of Blockchain and Data Security（区块链与数据安全国家重点实验室）；Hangzhou High-Tech Zone (Binjiang) Institute of Blockchain and Data Security（杭州高科技园区（滨江）区块链与数据安全研究院）；Sun Yat-sen University（中山大学）；KAUST（卡塔尔大学）
SafeThinker: Reasoning about Risk to Deepen Safety Beyond Shallow Alignment
Nanjing University of Aeronautics and Astronautics（南京航空航天大学）；Didi International Business Group（滴滴国际商务集团）；Institute for AI, Tsinghua University（清华大学人工智能研究院）；Nanjing University of Information Science & Technology（南京信息科学技术大学）
LookAhead Tuning: Safer Language Models via Partial Answer Previews
Zhejiang University（浙江大学）；Zhejiang University - Ant Group Joint Laboratory of Knowledge Graph（浙江大学-蚂蚁集团知识图谱联合实验室）；Ant Group（蚂蚁集团）；National University of Singapore（新加坡国立大学）
SafeRL-Lite: A Lightweight, Explainable, and Constrained Reinforcement Learning Library
Vision Mentors Ltd.（愿景导师有限公司）；School of Computer Science & AI, SR University（计算机科学与人工智能学院，SR大学）；Department of Computer Science, MANIT（计算机科学系，MANIT）；University of West London（西伦敦大学）
