大模型对齐与安全

Anchors in the Machine: Behavioral and Attributional Evidence of Anchoring Bias in LLMs

Felipe Valencia-Clavijo

机构 * Dataplicada 纠错

专题命中 AI治理与伦理 ：safety(abstract)；分类 cs.CL、cs.AI

Matchings Under Biased and Correlated Evaluations

Amit Kumar, Nisheeth K. Vishnoi

专题命中 AI治理与伦理 ：alignment(abstract)；分类 cs.CY

Comments To appear in NeurIPS 2025

Information Science Principles of Machine Learning: A Causal Chain Meta-Framework Based on Formalized Information Mapping

Jianfeng Xu

机构 * Koguan School of Law, China Institute for Smart Justice, School of Computer Science, Shanghai Jiao Tong University（柯 guar 法学院、中国智能正义研究院、计算机科学学院、上海交通大学） 纠错

专题命中 AI治理与伦理 ：safety(abstract)；分类 cs.AI

Investigating Prosocial Behavior Theory in LLM Agents under Policy-Induced Inequities

Yujia Zhou, Hexi Wang, Qingyao Ai, Zhen Wu, Yiqun Liu

专题命中 AI治理与伦理 ：alignment(abstract)

Comments Accepted by AAAI2026

Zeroth-Order Adaptive Neuron Alignment Based Pruning without Re-Training

Elia Cunegatti, Leonardo Lucio Custode, Giovanni Iacca

机构 * University of Trento（特伦托大学） 纠错

专题命中 其他安全 ：alignment(title,abstract)；分类 cs.CL、cs.AI、cs.LG

Comments Published in Transactions on Machine Learning Research (TMLR)

Data Trajectory Alignment for LLM Domain Adaptation: A Two-Phase Synthesis Framework for Telecommunications Mathematics

Zhicheng Zhou, Jing Li, Suming Qiu, Junjie Huang, Linyuan Qiu, Zhijie Sun

机构 * Global Technical Service (GTS) Huawei Technologies Co., Ltd（华为技术有限公司全球技术服务部） 纠错

专题命中 其他安全 ：alignment(title,abstract)；分类 cs.AI、cs.LG

DiA-gnostic VLVAE: Disentangled Alignment-Constrained Vision Language Variational AutoEncoder for Robust Radiology Reporting with Missing Modalities

Nagur Shareef Shaik, Teja Krishna Cherukuri, Adnan Masood, Dong Hye Ye

专题命中 其他安全 ：alignment(title,abstract)；分类 cs.AI、cs.LG

Comments Accepted for Oral Presentation at the 40th AAAI Conference on Artificial Intelligence (AAAI-26), Main Technical Track

Dynamics-Decoupled Trajectory Alignment for Sim-to-Real Transfer in Reinforcement Learning for Autonomous Driving

Thomas Steinecker, Alexander Bienemann, Denis Trescher, Thorsten Luettel, Mirko Maehlisch

机构 * University of the Bundeswehr Munich（联邦国防军慕尼黑大学） 纠错

专题命中 其他安全 ：alignment(title,abstract)；分类 cs.LG

HiMo-CLIP: Modeling Semantic Hierarchy and Monotonicity in Vision-Language Alignment

Ruijia Wu, Ping Chen, Fei Shen, Shaoan Zhao, Qiang Hui, Huanlin Gao, Ting Lu, Zhaoxiang Liu, Fang Zhao, Kai Wang, Shiguo Lian

专题命中 其他安全 ：alignment(title,abstract)；分类 cs.CL

Comments Accepted by AAAI 2026 as an Oral Presentation (13 pages, 7 figures, 7 tables)

Journal ref AAAI2026

Approximating the Mathematical Structure of Psychodynamics

Bryce-Allen Bagley, Navin Khoshnan

机构 * Stanford University（斯坦福大学） ； Mathematical Medicine Group（数学医学组） ； Department of Neurosurgery（神经外科系） ； Physician-Scientist Training Program（医师科学家培训计划） 纠错

专题命中 其他安全 ：alignment(abstract)；safety(abstract)；AI safety(abstract)；分类 cs.CL、cs.CY

Temporal Alignment of Time Sensitive Facts with Activation Engineering

Sanjay Govindan, Maurice Pagnucco, Yang Song

机构 * University of New South Wales（新南威尔士大学） 纠错

专题命中 其他安全 ：alignment(title)；分类 cs.CL、cs.LG

AdaRec: Adaptive Recommendation with LLMs via Narrative Profiling and Dual-Channel Reasoning

Meiyun Wang, Charin Polpanumas

机构 * The University of Tokyo（东京大学） ； Amazon.com, Inc.（亚马逊公司） 纠错

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI

Wasm: A Pipeline for Constructing Structured Arabic Interleaved Multimodal Corpora

Khalil Hennara, Ahmad Bastati, Muhammad Hreden, Mohamed Motasim Hamed, Zeina Aldallal, Sara Chrouf, Safwan AlModhayan

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI

Rethinking what Matters: Effective and Robust Multilingual Realignment for Low-Resource Languages

Quang Phuoc Nguyen, David Anugraha, Felix Gaschi, Jun Bin Cheng, En-Shiun Annie Lee

机构 * Ontario Tech University（安大略技术大学） ； Stanford University（斯坦福大学） ； SAS Posos ； University of Toronto（多伦多大学） 纠错

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI

Comments Accepted to IJCNLP-AACL 2025

Explicit Knowledge-Guided In-Context Learning for Early Detection of Alzheimer's Disease

Puzhen Su, Yongzhu Miao, Chunxi Guo, Jintao Tang, Shasha Li, Ting Wang

机构 * College of Computer Science and Technology, National University of Defense Technology（计算机科学与技术学院，国防科技大学） 纠错

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI

Comments This paper was accepted by IEEE BIBM 2025 conference

Do Street View Imagery and Public Participation GIS align: Comparative Analysis of Urban Attractiveness

Milad Malekzadeh, Elias Willberg, Jussi Torkko, Silviya Korpilo, Kamyar Hasanzadeh, Olle Järv, Tuuli Toivonen

专题命中 其他安全 ：alignment(abstract)；分类 cs.CY、cs.LG

Urban Computing in the Era of Large Language Models

Zhonghang Li, Lianghao Xia, Xubin Ren, Jiabin Tang, Tianyi Chen, Yong Xu, Chao Huang

机构 * South China University of Technology（华南理工大学） ； The University of Hong Kong（香港大学） ； City University of Hong Kong（城市大学） 纠错

专题命中 其他安全 ：safety(abstract)；分类 cs.CL、cs.CY

Comments https://github.com/HKUDS/Awesome-LLM4Urban-Papers

Journal ref ACM TIST, 2025

Revealing emergent human-like conceptual representations from language prediction

Ningyu Xu, Qi Zhang, Chao Du, Qiang Luo, Xipeng Qiu, Xuanjing Huang, Menghan Zhang

机构 * College of Computer Science and Artificial Intelligence（计算机科学与人工智能学院） ； Fudan University（复旦大学） ； Institute of Modern Languages and Linguistics（现代语言与语言学研究所） ； Shanghai Key Laboratory of Intelligent Information Processing（上海智能信息处理重点实验室） ； Research Institute of Intelligent Complex Systems（智能复杂系统研究所） ； Institute of Trustworthy Embodied Artificial Intelligence（可信具身人工智能研究所） ； State Key Laboratory of Genetics and Development of Complex Phenotypes（复杂表型遗传与发育国家重点实验室） 纠错

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL、cs.AI

Comments 66 pages. Accepted manuscript. Final version published in Proceedings of the National Academy of Sciences (PNAS): https://www.pnas.org/doi/10.1073/pnas.2512514122

Journal ref Proceedings of the National Academy of Sciences, U.S.A., 122 (44) e2512514122 (2025)

Leveraging Text-Driven Semantic Variation for Robust OOD Segmentation

Seungheon Song, Jaekoo Lee

专题命中 其他安全 ：safety(abstract)；分类 cs.AI

Comments 8 pages, 5 figure references, 2025 IEEE/RSJ International Conference on Intelligent Robots and Systems (IROS) submission

LLM Driven Processes to Foster Explainable AI

Marcel Pehlke, Marc Jansen

机构 * University of Applied Sciences Ruhr West（鲁尔西部应用科学大学） 纠错

专题命中 其他安全 ：alignment(abstract)；分类 cs.AI

LLM For Loop Invariant Generation and Fixing: How Far Are We?

Mostafijur Rahman Akhond, Saikat Chakraborty, Gias Uddin

机构 * York University（约克大学） ； Microsoft Research（微软研究院） 纠错

专题命中 其他安全 ：safety(abstract)；分类 cs.AI

Comments This work has been submitted to the IEEE for possible publication

GlitchMiner: Mining Glitch Tokens in Large Language Models via Gradient-based Discrete Optimization

Zihui Wu, Haichang Gao, Ping Wang, Shudong Zhang, Zhaoxiang Liu, Shiguo Lian

专题命中 其他安全 ：safety(abstract)；分类 cs.AI

Lived Experience in Dialogue: Co-designing Personalization in Large Language Models to Support Youth Mental Well-being

Kathleen W. Guan, Sarthak Giri, Mohammed Amara, Bernard J. Jansen, Enrico Liscio, Milena Esherick, Mohammed Al Owayyed, Ausrine Ratkute, Gayane Sedrakyan, Mark de Reuver, Joao Fernando Ferreira Goncalves, Caroline A. Figueroa

专题命中 其他安全 ：alignment(abstract)；分类 cs.AI

SMAGDi: Socratic Multi Agent Interaction Graph Distillation for Efficient High Accuracy Reasoning

Aayush Aluru, Myra Malik, Samarth Patankar, Spencer Kim, Kevin Zhu, Sean O'Brien, Vasu Sharma

机构 * Algoverse AI Research（Algoverse AI研究） 纠错

专题命中 其他安全 ：alignment(abstract)；分类 cs.AI

Comments Multi-Turn Interactions in Large Language Models (MTI-LLM) Workshop at NeurIPS 2025

OPLoRA: Orthogonal Projection LoRA Prevents Catastrophic Forgetting during Parameter-Efficient Fine-Tuning

Yifeng Xiong, Xiaohui Xie

专题命中 其他安全 ：alignment(abstract)；分类 cs.CL

CAMP-VQA: Caption-Embedded Multimodal Perception for No-Reference Quality Assessment of Compressed Video

Xinyi Wang, Angeliki Katsenou, Junxiao Shen, David Bull

机构 * School of Computer Science, University of Bristol（布里斯托大学计算机科学学院） 纠错

专题命中 其他安全 ：alignment(abstract)

Comments 14 pages, 6 figures

F2GAN: A Feature-Feedback Generative Framework for Reliable AI-Based Fault Diagnosis in Inverter-Dominated Microgrids

Swetha Rani Kasimalla, Kuchan Park, Junho Hong, Young-Jin Kim

专题命中 其他安全 ：alignment(abstract)

AGO: Adaptive Grounding for Open World 3D Occupancy Prediction

Peizheng Li, Shuxiao Ding, You Zhou, Qingwen Zhang, Onat Inak, Larissa Triess, Niklas Hanselmann, Marius Cordts, Andreas Zell

机构 * Mercedes-Benz AG（梅赛德斯-奔驰集团） ； University of Tübingen（图宾根大学） ； Tübingen AI Center（图宾根人工智能中心） ； University of Bonn（波恩大学） ； RPL ； KTH Royal Institute of Technology（皇家理工学院） ； TU Berlin（柏林技术大学） 纠错

专题命中 其他安全 ：alignment(abstract)

LangBridge: Interpreting Image as a Combination of Language Embeddings

Jiaqi Liao, Yuwei Niu, Fanqing Meng, Hao Li, Changyao Tian, Yinuo Du, Yuwen Xiong, Dianqi Li, Xizhou Zhu, Li Yuan, Jifeng Dai, Yu Cheng

机构 * Shanghai AI Laboratory（上海人工智能实验室） ； The Chinese University of Hong Kong（香港中文大学） ； Tsinghua University（清华大学） ； SenseTime Research（商汤科技研究院） ； Shanghai Jiao Tong University（上海交通大学） ； Peking University（北京大学） ； PengCheng Laboratory（鹏城实验室） ； Chongqing University（重庆大学） 纠错

专题命中 其他安全 ：alignment(abstract)

Comments The code and weights are open-sourced. Project page: https://curryx-001.github.io/LangBridge.github.io/

Taming Multi-Domain, -Fidelity Data: Towards Foundation Models for Atomistic Scale Simulations

Tomoya Shiota, Kenji Ishihara, Tuan Minh Do, Toshio Mori, Wataru Mizukami

专题命中 其他安全 ：alignment(abstract)