Translation in progress Some entries still include text in the original language.
Oct 12 · Monday
All papers Found 12 papers Another 265 papers do not have category tags yet and are not included in filtered results.
“
About this collection Collection in progress We are building a large AI safety paper collection to help researchers find work related to their interests. Papers are collected automatically andfiltered only for relevance to AI safety , without judging quality, novelty or reading value. The default “Annotated first” shows papers with tags or published paper analyses first. Each group is ordered by original publication date, newest first, with undated papers last. “Newest first” uses publication dates only. This order is not a quality ranking.
One-line summaries, tags and paper analyses are prepared by AI from the papers. Language models may have systematic biases and produce hallucinations.We plan to carry out human annotation and review to reduce errors in this content.
Analyzing why optimizing teaching adaptation metrics collapses tutoring choices and diverges from expert judgment
Published Oct 8
Evaluated models Gemini 3.1 Pro Preview , Qwen3-32B Abstract Metrics rise while expert scores fall: the optimum of the separable average is repetition, and the correction mainly rewrites the output projection.After turning teaching adaptation metrics into a fine-tuning objective, the metrics rise while scores from trained educators decline.
测量财务事实变少时身份对股权配置建议的替代
Published Oct 6
Evaluated models Llama-3.1-8B-Instruct , Qwen2.5-1.5B-Instruct Abstract 身份摆幅随披露减少上升。同一财务只改人设，测量披露变薄时股权建议里的身份替代。
提出 LineupRL，用描述到序列识别作可验证奖励训练时序描述模型
Published Oct 1 Abstract LineupRL 用统计近邻上的序列识别做可验证奖励，3B 描述器超过 SFT、两种 RL 奖励和 72B 教师。LineupRL 是给开放式时间序列描述用的 RLVR：不模仿合成参考，而奖励一条描述能否让冻结文本 LLM 从相近序列里认出目标。
提出 MASCRDM，在监督微调中实时检测并缓解偏见
Published Sep 30
Evaluated models Qwen3-8B-Base , Llama-3.1-8B Abstract MASCRDM 在 SFT 中按法规图谱检测数据、结构与损失风险，BBQ 总准确率在两模型上最高。MASCRDM 把 AI 法律规章编成知识图谱，再由五个智能体在 LLM 训练中做实时合规检测与缓解，实验以偏见与歧视为代表风险。
提出可解释 rubric 奖励，将宪法转为可调权重并用于监督微调
Published Sep 27
Evaluated models Gemma 3 12B , Qwen3 14B , GPT-OSS 20B Abstract 宪法级可解释奖励让优先级可检查、可改写，并带入训练后的行为。作者给出一条从通用宪法到可调训练奖励的路径，用来检查并改写模型实际被优化的优先级。
B IAS R EDUCER 只改奖励头，按数据集选择属性编辑。
Published Sep 26
Evaluated models Skywork-Reward-V2-Qwen3-1.7B , GRM-Llama3.2-3B-rewardmodel-ft , RM-Mistral-7B and others7 Abstract BIASREDUCER 按数据集选择奖励头编辑，五模型三基准平均提升并迁移到下游训练。BIASREDUCER 是一种只改线性奖励头的奖励模型编辑框架，用来降低模型对预定义表面属性的依赖，并按新数据集决定改哪些属性。
用 CDPG 与 BTM 追踪思考如何改变反事实公平翻转
Published Sep 25
Evaluated models QwQ-32B , DeepSeek-R1-Distill-Qwen-32B , Qwen3-32B and others5 审计视觉语言模型在敏感图像上从拒答转向国家立场重构的行为
Published Aug 12
Evaluated models Qwen2-VL-7B , Qwen2.5-VL-7B , Qwen3-VL-8B and others9 Abstract 论文系统审计了 VLM 多模态审查，发现审查正从显式拒答转向隐蔽重构，且中文提示与文本先验会放大重构倾向。论文系统研究了中外开源视觉语言模型在政治敏感图像上的审查行为，主要内容如下。
测量内容保留的修辞改写如何改变 AI 审稿分数
Published Aug 10
Evaluated models Gemini 3.5 Flash-Lite , Qwen 3.5 Flash , GPT-5 mini and others5 Abstract 作者称修辞影响选择性且依赖配置，证据与新颖性反差大，复杂流程不保证更大增益。作者分析内容保留的修辞变化如何改变 AI 科学评审，并把这种现象称为对 AI 评审者的潜在奖励投机。
提出 BPA 改写 profile 与记忆，重新诱发智能体对人类的外群体偏向
Published Jan 1
Evaluated models gpt-4o-mini Abstract 最小群体线索下出现内群体偏向。作者考察 LLM 智能体在任意群体边界下会不会把人类当外群体，并用信念投毒检验面向人类的抑制是否依赖身份信念。
用反事实与嵌入距离评测作文反馈的性别偏差
Published Nov 11, 2025
Evaluated models GPT-5 mini , GPT-4o mini , DeepSeek-R1 and others6 Abstract 六种 LLM 作文反馈对隐式性别替换不对称，显式线索只让部分模型的语义距离变大。作者用句嵌入距离和置换检验，给教育场景中的 LLM 性别偏见审计做一个可计算基准，案例是学生议论文的形成性反馈。
用 SPR 评测 AI 科学家系统的方法学缺陷
Published Sep 11, 2025
Evaluated models gemini-2.5-flash-preview-05-20 Abstract 作者用 SPR 诊断两个开源系统的四种方法学缺陷，并建议审稿时提交日志与代码。作者诊断两个开源系统的全自动科研流程，看方法学缺陷会不会留在终稿看不见的决策里。
You have reached the end