1
Selective Safety Steering via Value-Filtered Decoding
提出选择性安全引导方法，通过价值过滤解码提升LLM安全性
arXiv:2605.14746v2 Announce Type: replace Abstract: While large language models (LLMs) are trained to align with human values, their generations may s…
2
On the Convergence of Self-Improving Online LLM Alignment
揭示大语言模型在线自我改进对齐的收敛性理论，为LLM安全与性能优化提供数学基础，UAI 2026收录。
arXiv:2606.31524v1 Announce Type: cross Abstract: The Self-Improving Alignment (SAIL) algorithm addresses distribution shift by reducing a bilevel for…
3
Towards Spec Learning: Inference-Time Alignment from Preference Pairs
通过偏好对进行推理时对齐，提出高效方案避免昂贵微调
arXiv:2606.24004v1 Announce Type: cross Abstract: Steering a large language model (LLM) toward a desired behavior typically relies on an iterative pro…
4
Societal Alignment Frameworks Can Improve LLM Alignment
引入社会对齐框架，为提升大语言模型对齐效果开辟新路径，论文值得AI从业者细读。
arXiv:2503.00069v2 Announce Type: replace-cross Abstract: Recent progress in large language models (LLMs) has focused on producing responses that meet…
5
Certifiable Safe RLHF: Semantic Grounding and Fixed Penalty Constraint Optimization for Safer LLM Alignment
语义基础+固定惩罚约束优化，让大模型对齐过程获得可认证的安全保障
arXiv:2510.03520v2 Announce Type: replace-cross Abstract: Ensuring safety is a foundational requirement for large language models (LLMs). Achieving an…
6
ALIGNBEAM : Inference-Time Alignment Transfer via Cross-Vocabulary Logit Mixing
提出推理时对齐迁移新方法，跨词汇混合logit无需额外训练即可转移对齐能力，为大模型高效部署提供新思路。
arXiv:2606.12342v1 Announce Type: cross Abstract: Domain fine-tuning degrades the safety of large language models: fine-tuned specialists readily comp…
7
Korean Culture into LLM Alignment: Toward Cultural Coherence
将韩国文化系统融入大模型对齐，推动跨文化一致性研究，被ICML 2026研讨会收录的学术论文。
arXiv:2606.06797v1 Announce Type: new Abstract: Cultural-aspect work on large language models is dominated by a negative target: which outputs to supp…
8
TriAlign: Towards Universal Truth Consistency in Personalized LLM Alignment
提出TriAlign新框架，解决大模型个性化对齐中的真理一致性问题，思路新颖。
arXiv:2606.01755v1 Announce Type: new Abstract: Personalized large language models adapt responses to users' preferences and social attributes, but ca…
9
InfiMed-ORBIT: Aligning LLMs on Open-Ended Complex Tasks via Rubric-Based Incremental Training
用评分标准做增量训练，让大模型在开放式医疗对话中也能高效对齐，突破传统RL局限。
arXiv:2510.15859v4 Announce Type: replace-cross Abstract: Reinforcement learning (RL) has driven recent breakthroughs in large language models (LLMs),…
10
Sparse Autoencoders Reveal Cortical Brain-LLM Semantic Mapping
稀疏自编码器首次建立大脑皮层语义拓扑与LLM特征的可解释映射，跨学科新突破
Article URL: https://letsdatascience.com/news/sparse-autoencoders-reveal-cortical-brain-llm-semantic-mappi-bc586635 Comments URL: https://news.ycombin…
11
Whose Alignment? Comparing LLM Process Alignment Across Diverse Organizational Decision Contexts
探究组织场景下大语言模型的过程对齐，如何超越个人偏见反映真实政策。
arXiv:2605.25256v1 Announce Type: new Abstract: Aligning AI systems with organizational decision-making is typically framed as a single-target problem…
12
Convex Optimization for Alignment and Preference Learning on a Single GPU
单GPU实现凸优化方法，高效解决LLM偏好对齐难题，降低RLHF计算成本。
arXiv:2605.23244v1 Announce Type: new Abstract: Fine-tuning large language models (LLMs) to align with human preferences has driven the success of sys…
13
Sparse Autoencoders Map Brain-LLM Alignment onto Cortical Semantic Topography
揭秘稀疏自编码器如何揭示大脑与LLM的语义对齐模式，探索皮层拓扑结构中的智能映射
arXiv:2605.23035v1 Announce Type: cross Abstract: Intermediate layers of large language models (LLMs) best predict human brain responses to language, …
14
Brain-LLM Alignment Tracks Training Data, Not Typology
跨语言验证发现大脑语言网络与LLM的对齐主要受训练数据驱动，而非语言类型学差异。
arXiv:2605.23032v1 Announce Type: cross Abstract: Brain-LLM alignment is well established in English, yet the brain's language network is neuroanatomi…
15
Benchmarking and Improving Monitors for Out-Of-Distribution Alignment Failure in LLMs
针对LLM在分布外场景下的对齐失败问题，提出基准测试和改进监测器的新方法，提升AI安全可靠性。
arXiv:2605.21602v1 Announce Type: new Abstract: Many safety and alignment failures of large language models (LLMs) occur due to out-of-distribution (O…
16
From Parameters to Data: A Task-Parameter-Guided Fine-Tuning Pipeline for Efficient LLM Alignment
提出强映射假设，揭示数据选择与参数高效微调的内在耦合，为LLM对齐提供新思路。
arXiv:2605.21558v1 Announce Type: cross Abstract: Adapting Large Language Models (LLMs) to specialized domains typically incurs high data and computat…
17
We Think, Therefore We Align LLMs to Helpful, Harmless and Honest Before They Go Wrong
LLM对齐新思路：在模型出错前就教会它思考“有用、无害、诚实”
arXiv:2509.22510v3 Announce Type: replace Abstract: Alignment of Large Language Models (LLMs) is the ability to satisfy desired objectives during gene…
18
Not Just RLHF: Why Alignment Alone Won't Fix Multi-Agent Sycophancy
研究发现多智能体系统在同伴分歧下“屈服”并非RLHF特有，基础模型同样存在该漏洞，挑战了传统对齐认知。
arXiv:2605.12991v2 Announce Type: replace Abstract: LLM-based multi-agent pipelines flip from correct to incorrect answers under simulated peer disagr…
19
S2Aligner: Pair-Efficient and Transferable Pre-Training for Sparse Text-Attributed Graphs
提出针对稀疏文本属性图的高效可迁移预训练方法S2Aligner，解决LLM对齐中的监督不足问题。
arXiv:2605.18579v1 Announce Type: new Abstract: Pre-training on text-attributed graphs (TAGs) is central to building transferable graph foundation mod…