大模型对齐与安全 - arXivDaily 专题

2512.04334 2026-02-20 cs.HC 50%

Human-controllable AI: Meaningful Human Control

可受人类控制的AI：有意义的人类控制

Chengke Liu, Wei Xu

专题命中 AI治理与伦理：alignment(abstract)

AI总结本文提出有意义的人类控制（MHC）作为AI发展的重要原则，强调技术设计、治理和人类共同作用的重要性，旨在实现人类中心的AI发展。

Comments 52 pages

2602.09369 2026-02-13 cs.CR 50%

Timing and Memory Telemetry on GPUs for AI Governance

GPU上的定时和内存遥测用于AI治理

Saleh K. Monfared, Fatemeh Ganji, Dan Holcomb, Shahin Tajik

专题命中 AI治理与伦理：alignment(abstract)

AI总结通过GPU定时和内存遥测技术，实现对大规模AI计算资源的治理，以确保资源使用符合政策和法律要求。

2602.07312 2026-02-10 cs.HC 50%

Navigating Algorithmic Opacity: Folk Theories and User Agency in Semi-Autonomous Vehicles

穿越算法隐匿性：半自动驾驶车辆中用户自主性的folk理论

Yehuda Perry, Tawfiq Ammari

专题命中 AI治理与伦理：safety(abstract)

AI总结研究探讨了半自动驾驶车辆司机如何通过folk理论理解算法行为，揭示了司机在算法治理中的被动地位，并提出参与式算法治理框架以提升透明度和用户参与度。

2602.03155 2026-02-04 cs.HC 50%

Is It Possible to Make Chatbots Virtuous? Investigating a Virtue-Based Design Methodology Applied to LLMs

能否使聊天机器人变得有德性？探讨一种基于德性的设计方法应用于大语言模型

Matthew P. Lad, Louisa Conwill, Megan Levis Scheirer

专题命中 AI治理与伦理：safety(abstract)

AI总结本文探讨了基于德性的设计方法在大语言模型中的应用，通过伦理设计模式的编目和评估，提出了一种提升LLM伦理性的设计框架，并指出其在准确性和安全性方面的优势及潜在实施挑战。

2601.10220 2026-01-16 cs.SE 50%

Agentic Pipelines in Embedded Software Engineering: Emerging Practices and Challenges

嵌入式软件工程中的代理流水线：新兴实践与挑战

Simin Sun, Miroslaw Staron

专题命中 AI治理与伦理：safety(abstract)

AI总结本文探讨嵌入式软件工程中生成式AI的整合挑战与实践，通过专家访谈揭示代理流水线的发展与可持续采用策略。

2601.07051 2026-01-13 cs.SE 50%

Between Policy and Practice: GenAI Adoption in Agile Software Development Teams

在政策与实践之间：生成式AI在敏捷软件开发团队中的采用

Michael Neumann, Lasse Bischof, Nic Elias Hinz, Luca Stockmann, Dennis Schrader, Ana Carolina Ahaus, Erim Can Demirci, Benjamin Gabel, Maria Rauschenberger, Philipp Diebold, Henning Fritzemeier, Adam Przybylek

专题命中 AI治理与伦理：alignment(abstract)

AI总结本研究探讨了生成式AI在敏捷软件开发团队中的实际采用情况，发现其主要用于创意任务和文档协助，但面临数据隐私和治理等障碍，需通过TOE框架协调政策与实践以实现有效整合。

2601.05555 2026-01-12 cs.SE 50%

An Empirical Study of Policy-as-Code Adoption in Open-Source Software Projects

对开源软件项目中政策即代码采用的实证研究

Patrick Loic Foalem, Foutse Khomh, Leuson Da Silva, Ettore Merlo

专题命中 AI治理与伦理：trustworthy(abstract)

AI总结本文通过分析399个开源项目，揭示了政策即代码工具在治理、配置控制和文档中的广泛应用，并提出了包含5类15子类的分类法，为未来研究提供实证基础。

2510.00762 2026-01-05 cs.SE cs.HC 50%

AI Where It Matters: Where, Why, and How Developers Want AI Support in Daily Work

AI 何时重要：开发者在日常工作中需要 AI 支持的地点、原因和方式

Rudrajit Choudhuri, Carmen Badea, Christian Bird, Jenna Butler, Rob DeLine, Brian Houck

专题命中 AI治理与伦理：alignment(abstract)

AI总结本文通过研究开发者在日常工作中对AI支持的需求，揭示了AI在不同任务中的应用模式及负责任AI的优先事项。

Comments ICSE-SEIP'26

2506.22876 2025-12-23 cs.MA 50%

Cooperation as a Black Box: Conceptual Fluctuation and Diagnostic Tools for Misalignment in MAS

协作作为黑箱：多智能体系统中概念波动与对齐偏差的诊断工具

Shayak Nandi, Fernanda M. Eliott

专题命中 AI治理与伦理：alignment(abstract)

AI总结本文提出马赛克框架，用于诊断多智能体系统中概念波动与对齐偏差，强调术语一致性与道德基础以确保系统技术与道德对齐。

2512.18234 2025-12-23 cs.HC 50%

The Social Blindspot in Human-AI Collaboration: How Undetected AI Personas Reshape Team Dynamics

人机协作中的社会盲区：未被识别的人工智能角色如何重塑团队动态

Lixiang Yan, Xibin Han, Yu Zhang, Samuel Greiff, Inge Molenaar, Roberto Martinez-Maldonado, Yizhou Fan, Linxuan Zhao, Xinyu Li, Yueqiao Jin, Dragan Gašević

专题命中 AI治理与伦理：safety(abstract)

AI总结研究发现AI通过人格设定影响团队协作，即使用户未察觉其存在，揭示了社会盲区现象。

2512.08374 2025-12-10 cs.CV 50%

The Unseen Bias: How Norm Discrepancy in Pre-Norm MLLMs Leads to Visual Information Loss

看不见的偏见：预规范MLLM中规范差异如何导致视觉信息丢失

Bozhou Li, Xinda Xue, Sihan Yang, Yang Shi, Xinlong Chen, Yushuo Guan, Yuanxing Zhang, Wentao Zhang

机构 * Peking University（北京大学）； School of Artificial Intelligence, University of Chinese Academy of Sciences（中国科学院大学人工智能学院）； Xi’an Jiaotong University（西安交通大学）； Kling Team, Kuaishou Technology（快手科技 Kling 团队）

专题命中 AI治理与伦理：alignment(abstract)

AI总结本文揭示了预规范MLLM中规范差异导致的视觉信息丢失问题，并提出通过插入层规范层来解决这一问题，提升模型整体能力。

2512.03439 2025-12-04 cs.IR 50%

LLM as Explainable Re-Ranker for Recommendation System

大语言模型作为可解释的重排序器用于推荐系统

Yaqi Wang, Haojia Sun, Shuting Zhang

专题命中 AI治理与伦理：alignment(abstract)

AI总结本文提出利用大语言模型作为可解释的重排序器，结合传统推荐模型提升推荐系统的准确性和可解释性。

2512.03180 2025-12-04 cs.MA cs.ET 50%

AGENTSAFE: A Unified Framework for Ethical Assurance and Governance in Agentic AI

AGENTSAFE：面向代理AI的统一伦理保障与治理框架

Rafflesia Khan, Declan Joyce, Mansura Habiba

专题命中 AI治理与伦理：safety(abstract)

AI总结 AGENTSAFE提出一个统一的治理框架，通过风险分类转化为可操作的设计、运行时和审计控制，提供可衡量的预部署保障，并通过运行时治理和问责机制建立代理AI生态系统的信任。

Comments 12 pages, 1 figure

2512.02562 2025-12-03 eess.SY cs.SY 50%

Intervention Strategies for Fairness and Efficiency at Autonomous Single-Intersection Traffic Flows

自主单交叉口交通流的公平性与效率干预策略

Salman Ghori, Ania Adil, Melkior Ornik, Eric Feron

专题命中 AI治理与伦理：safety(abstract)

AI总结本文提出了一种基于混合整数线性规划的干预策略，用于在无信号灯的交叉口上优化自主代理的公平性与效率，通过仿真验证早干预和公平性意识控制的效果。

2512.01571 2025-12-02 cs.NI 50%

Velocity-Adaptive Access Scheme for Semantic-Aware Vehicular Networks: Joint Fairness and AoI Optimization

面向语义感知车联网的自适应接入方案：公平性与信息年龄优化的联合优化

Xiao Xu, Qiong Wu, Pingyi Fan, Kezhi Wang, Nan Cheng, Wen Chen, Khaled B. Letaief

专题命中 AI治理与伦理：safety(abstract)

AI总结本文提出一种面向语义感知车联网的自适应接入方案，通过联合优化公平性与信息年龄，利用顺序凸近似和大语言模型算法提升通信效率与安全性。

Comments This paper has been submitted to IEEE transactions on moblie computing

2512.00428 2025-12-02 cs.CV 50%

Recognizing Pneumonia in Real-World Chest X-rays with a Classifier Trained with Images Synthetically Generated by Nano Banana

用Nano Banana生成的合成图像训练分类器以在真实世界胸片中识别肺炎

Jiachuan Peng, Kyle Lam, Jianing Qiu

机构 * Mohamed bin Zayed University of Artificial Intelligence（穆罕默德·本·扎耶德人工智能大学）； University of Oxford（牛津大学）； Imperial College London（伦敦帝国学院）

专题命中 AI治理与伦理：alignment(abstract)

AI总结利用Nano Banana生成的合成图像训练分类器，实现在真实世界胸片中识别肺炎，取得较高AUROC和AUPR，但存在提示设计和数据对齐的挑战。

Comments 9 pages

2511.21579 2025-12-01 cs.CV 50%

Harmony: Harmonizing Audio and Video Generation through Cross-Task Synergy

Harmony: 通过跨任务协同实现音频和视频生成

Teng Hu, Zhentao Yu, Guozhen Zhang, Zihan Su, Zhengguang Zhou, Youliang Zhang, Yuan Zhou, Qinglin Lu, Ran Yi

机构 * Shanghai Jiao Tong University（上海交通大学）； Tencent Hunyuan Project（腾讯文心项目）

专题命中 AI治理与伦理：alignment(abstract)

AI总结 Harmony通过跨任务协同机制和同步增强CFG，实现高效音频视频同步生成。

2511.19886 2025-11-26 cs.CR cs.CV 50%

Frequency Bias Matters: Diving into Robust and Generalized Deep Image Forgery Detection

频率偏差至关重要：深入探讨鲁棒且通用的深度图像伪造检测

Chi Liu, Tianqing Zhu, Wanlei Zhou, Wei Zhao

机构 * Faculty of Data Science, City University of Macau（数据科学学院，澳门城市大学）； Shenzhen Institute of Advanced Technology, Chinese Academy of Sciences（深圳先进技术研究所，中国科学院）

专题命中 AI治理与伦理：alignment(abstract)

AI总结本文从频率视角分析深度图像伪造检测器的通用性和鲁棒性问题，提出两步频率对齐方法以提升检测可靠性并增强反伪造能力。

Comments Accepted for publication in IEEE Transactions on Dependable and Secure Computing

2511.15859 2025-11-21 cs.SE 50%

RE for AI in Practice: Managing Data Annotation Requirements for AI Autonomous Driving Systems

在实践中为AI管理数据标注需求：为AI自主驾驶系统管理数据标注要求

Hina Saeeda, Mazen Mohamad, Eric Knauss, Jennifer Horkoff, Ali Nouri

专题命中 AI治理与伦理：safety(abstract)

AI总结本研究通过实证分析，探讨了自动驾驶系统中数据标注需求的定义、挑战及最佳实践，为提升AI系统可靠性提供指导。

2410.22896 2025-11-17 physics.soc-ph 50%

Modelling vehicle and pedestrian collective dynamics: Challenges and advances

Antoine Tordeux, Cécile Appert-Rolland, Alexandre Nicolas, Armin Seyfried, Denis Ullmo

专题命中 AI治理与伦理：safety(abstract)

Comments 20 pages, 6 figures

2507.14686 2025-11-12 cs.CV 50%

From Semantics, Scene to Instance-awareness: Distilling Foundation Model for Grounded Open-vocabulary Situation Recognition

Chen Cai, Tianyi Liu, Jianjun Gao, Wenyang Liu, Kejun Wu, Ruoyu Wang, Yi Wang, Soo Chin Liew

机构 * National University of Singapore（新加坡国立大学）； Nanyang Technological University（南洋理工大学）； Huazhong University of Science and Technology（华中科技大学）； The Hong Kong Polytechnic University（香港理工大学）

专题命中 AI治理与伦理：alignment(abstract)

2505.15857 2025-11-11 cs.SI 50%

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

2502.14886 2025-11-04 cs.CV 50%

Surgical Scene Understanding in the Era of Foundation AI Models: A Comprehensive Review

Ufaq Khan, Umair Nawaz, Adnan Qayyum, Shazad Ashraf, Yutong Xie, Muhammad Haris Khan, Muhammad Bilal, Junaid Qadir

机构 * MBZ University of AI（MBZ人工智能大学）； Hamad Bin Khalifa University（哈马德·本·卡西姆大学）； Birmingham City University（伯明翰城市大学）； Qatar University（卡塔尔大学）； University Hospitals Birmingham and University of Birmingham（伯明翰大学医院和伯明翰大学）

专题命中 AI治理与伦理：alignment(abstract)

2511.00026 2025-11-04 cs.RO 50%

Gen AI in Automotive: Applications, Challenges, and Opportunities with a Case study on In-Vehicle Experience

Chaitanya Shinde, Divya Garikapati

专题命中 AI治理与伦理：safety(abstract)

2408.08878 2025-10-23 cs.HC 50%

Knowledge Prompting: How Knowledge Engineers Use Large Language Models

Elisavet Koutsiana, Johanna Walker, Michelle Nwachukwu, Bohui Zhang, Albert Meroño-Peñuela, Elena Simperl

专题命中 AI治理与伦理：trustworthy(abstract)

2510.17842 2025-10-22 cs.SE cs.HC 50%

Vibe Coding: Toward an AI-Native Paradigm for Semantic and Intent-Driven Programming

Vinay Bamil

专题命中 AI治理与伦理：alignment(abstract)

Comments 10 pages, 1 figure, 2 tables

2510.17119 2025-10-21 cs.HC 50%

Design Framework for Conversational Agent in Couple relationships: A Systematic Review

Soyoung Jung, Sung Park

专题命中 AI治理与伦理：safety(abstract)

2405.07314 2025-10-21 cs.IR 50%

Learnable Item Tokenization for Generative Recommendation

Wenjie Wang, Honghui Bao, Xinyu Lin, Jizhi Zhang, Yongqi Li, Fuli Feng, See-Kiong Ng, Tat-Seng Chua

专题命中 AI治理与伦理：alignment(abstract)

Comments Accepted by CIKM 2024

2504.18837 2025-10-14 cs.SI 50%

Sentiment and Social Signals in the Climate Crisis: A Survey on Analyzing Social Media Responses to Extreme Weather Events

Pouya Shaeri, Yasaman Mohammadpour, Alimohammad Beigi, Ariane Middel

专题命中 AI治理与伦理：alignment(abstract)

Comments Accepted and Published in SBP-BRiMS 2025. 18th International Conference on Social Computing, Behavioral-Cultural Modeling & Prediction and Behavior Representation in Modeling and Simulation

2510.06556 2025-10-09 econ.TH 50%

Artificial Intelligence in Port Logistics: A Bibliometric Analysis of Technological Integration and Research Dynamics

Abdelhafid Khazzar, Yassine Sekaki, Yasser Lachhab, Said El-marzouki

专题命中 AI治理与伦理：safety(abstract)

1. AI治理与伦理 1852 篇

Human-controllable AI: Meaningful Human Control

Timing and Memory Telemetry on GPUs for AI Governance

Navigating Algorithmic Opacity: Folk Theories and User Agency in Semi-Autonomous Vehicles

Is It Possible to Make Chatbots Virtuous? Investigating a Virtue-Based Design Methodology Applied to LLMs

Agentic Pipelines in Embedded Software Engineering: Emerging Practices and Challenges

Between Policy and Practice: GenAI Adoption in Agile Software Development Teams

An Empirical Study of Policy-as-Code Adoption in Open-Source Software Projects

AI Where It Matters: Where, Why, and How Developers Want AI Support in Daily Work

Cooperation as a Black Box: Conceptual Fluctuation and Diagnostic Tools for Misalignment in MAS

The Social Blindspot in Human-AI Collaboration: How Undetected AI Personas Reshape Team Dynamics

The Unseen Bias: How Norm Discrepancy in Pre-Norm MLLMs Leads to Visual Information Loss

LLM as Explainable Re-Ranker for Recommendation System

AGENTSAFE: A Unified Framework for Ethical Assurance and Governance in Agentic AI

Intervention Strategies for Fairness and Efficiency at Autonomous Single-Intersection Traffic Flows

Velocity-Adaptive Access Scheme for Semantic-Aware Vehicular Networks: Joint Fairness and AoI Optimization

Recognizing Pneumonia in Real-World Chest X-rays with a Classifier Trained with Images Synthetically Generated by Nano Banana

Harmony: Harmonizing Audio and Video Generation through Cross-Task Synergy

Frequency Bias Matters: Diving into Robust and Generalized Deep Image Forgery Detection

RE for AI in Practice: Managing Data Annotation Requirements for AI Autonomous Driving Systems

Modelling vehicle and pedestrian collective dynamics: Challenges and advances

From Semantics, Scene to Instance-awareness: Distilling Foundation Model for Grounded Open-vocabulary Situation Recognition

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

Surgical Scene Understanding in the Era of Foundation AI Models: A Comprehensive Review

Gen AI in Automotive: Applications, Challenges, and Opportunities with a Case study on In-Vehicle Experience

Knowledge Prompting: How Knowledge Engineers Use Large Language Models

Vibe Coding: Toward an AI-Native Paradigm for Semantic and Intent-Driven Programming

Design Framework for Conversational Agent in Couple relationships: A Systematic Review

Learnable Item Tokenization for Generative Recommendation

Sentiment and Social Signals in the Climate Crisis: A Survey on Analyzing Social Media Responses to Extreme Weather Events

Artificial Intelligence in Port Logistics: A Bibliometric Analysis of Technological Integration and Research Dynamics