Translation in progress Some entries still include text in the original language.
Oct 13 · Tuesday
All papers Found 53 papers Another 215 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.
Auditing LLM judge reliability and defining trustworthy verdict rate and accuracy upper bound
Published Oct 8
Evaluated models GPT-5.2 , GPT-4o , Gemini-2.5-Pro and others6 Abstract The authors claim that comparing judges should look at T, not just accuracy.The authors audit whether LLM-as-a-Judge decisions can still be treated as actionable ground truth under repeated queries, swapped candidate order, and changed prompt formats.
提出包络采样,用少量真值标注重标定 LLM 评委以缓解奖励作弊
Published Oct 8
Evaluated models Qwen3-4B , Qwen3.8-27B , Qwen2.5-14B-Instruct Abstract 包络采样把真值标注打到代理尾部,同样预算下缓解 reward hacking。作者研究后训练前如何选择真值标注的查询分布,使重标定后的廉价代理更适合被优化。真奖励昂贵时,实践用 LLM judge 代替人类。若失准只发生在基座很少生成的输出上,on-policy 标注修不到这些点,随后的 KL 正则优化会把质量集中到被高估的输出。
提出 Persist-3 序贯检验,从隐藏表示识别未见越狱源
Published Oct 7
Evaluated models Llama2-7B-chat , Qwen2.5-7B-Instruct , Gemma-2-9B-it and others4 Abstract Persist-3 用有限窗口检验攻击流是否超出已知越狱源,检出率低于全历史 SR-DKW,但决策只依赖最近观测。Persist-3 是一个有限记忆的序贯检测器,用来判断攻击表示流是否已经不能由任何一个已知越狱源解释。 。
AI summary 研究者提出一种轻量级路由 logits 安全检测器,在提示词预填充阶段读取 MoE 视觉语言模型的专家路由信号,在生成前判断多模态请求是否安全,且不修改模型参数或专家路由。
Published Oct 6
Evaluated models Qwen3-VL-30B-A3B-Instruct , Kimi-VL-A3B-Instruct Abstract 读出而非操纵 router logits,可在不改模型的情况下降低两模型的组合式安全错误。作者在两个指令微调 MoE VLM 上比较改模型与读路由。组合式风险按 HoliSafe 分成五类:图文各自可安全,合在一起才不安全,或不安全内容只在图像里。
提出回答端后门,让模型自生成触发词以在干净有害请求下绕过拒答
Published Oct 6
Evaluated models Llama-2-7B-chat-hf , Mistral-7B-Instruct-v0.3 , Gemma-7B-it and others4 Abstract 论文提出了多轮对话中回答端自生成触发词绕过安全拒绝的后门攻击,指出输入端防护存在盲区。
评测文生图模型的有害内容生成能力与审核检测缺口
Published Oct 5
Evaluated models SDXL , Stable Diffusion 3.5 Large , FLUX.1-dev and others12 提出 VERA,审计 LLM 漏洞解释里的虚构推理
Published Oct 5
Evaluated models sonnet-5 , gemma-4-26b , llama-4-maverick and others8 提出 RubricArmor,用攻击–修复循环生成抗奖励作弊的评分准则
Published Oct 4
Evaluated models Qwen2.5-3B-Instruct , Qwen2.5-7B-Instruct , Qwen3-4B and others8 提出 PyINE 基准,训练捷径跟随模型并比较监督者
Published Oct 4
Evaluated models shortcut-following model organism , gpt-4o , gpt-5 and others12 Abstract PyINE 用可验证代码执行研究捷径失败的监督覆盖与成本权衡。PyINE 是用可执行程序研究推理模型引出与监督的框架,PyINE-v1 是其第一次发布。
提出 SecJev,用场景加权训练把决策模型适配为安全类型化判断
Published Oct 2
Evaluated models SecJev-0.8B , SecJev-2B , SecJev-4B and others4 Abstract SecJev 用共享打分器做安全决策,0.8B 的 task-macro 超过 Kev-9B。SecJev 是 0.8B 到 9B 的安全决策模型族,作者称据其所知是首个面向安全的 Jev-like 家族。
提出 MetaRubric,以证据感知奖励缓解评分标准强化学习的虚记分
Published Oct 2
Evaluated models Qwen3-4B , Qwen3-8B , Qwen3-VL and others5 Abstract MetaRubric 用证据约束并在训练中改 rubric,医学基准上相对静态 GRPO 有提高。MetaRubric 针对 rubric-based RL 里的空记分:评审在准则要求的信息或动作缺失时仍给高分。
系统评估循环语言模型加深 loop 后的 CoT 可监控性
Published Oct 2
Evaluated models Ouro-1.4B-Thinking , Ouro-2.6B-Thinking , Nanbeige4.2-3B and others4 Abstract 加深 loop 可在部分压力测试任务降低可监测性,循环架构本身并不必然更难监测。这项工作评估 LoopLM 的 CoT 可监测性:加深共享层的重复次数,以及“有循环结构”本身,会不会让决策关键因素更少出现在可监测文本里。
用分类后缀提升激活探针对分布外恶意输入的检测
Published Oct 2
Evaluated models Llama-3.1-8B , Qwen3.5-9B , Gemma-4-12B Abstract 作者称排序增益来自分类格式,低 FPR 精度来自点名准则。用户轮后的分类后缀,被作者用作激活探针的分布外检测干预:排序看格式,低 FPR 精度看是否点名恶意/良性。
检验低监控读数能否证明代码生成中的奖励作弊已受控
Published Oct 2
Evaluated models Llama-3.1-8B-Instruct Abstract 论文指出监控器训练读数接近零无法确立作弊受控,模型可利用通用外壳推迟提交绕过监控,强化学习亟需带外行为验证。本文是一篇针对强化学习可验证奖励后训练(RLVR)中监控器有效性与奖励作弊行为的实证分析研究。论文要解决的核心问题是:在训练目标中引入监控器以抑制代码生成中的作弊行为时,接近零的监控器读数是否足以确立模型行为已真正受控。
构建编码 RL 奖励作弊测试台 CATCH 并检验 CoT 监控失效
Published Sep 30
Evaluated models Qwen3-4B , Qwen3.5-27B Abstract CATCH 用双运行金标签研究编码 RL 的奖励黑客,并观察到监控会被注释适应削弱。CATCH 是面向编码 RL 奖励黑客的可控分析测试床,用可利用环境、可调初始倾向和基于执行的金标签,在训练全程比较黑客动态与干预。
提出 ProRubric,把加性评分聚合改为合取协议以抑制奖励作弊
Published Sep 30
Evaluated models Qwen3-4B , Qwen3-8B Abstract ProRubric 用离线合取维度替换加性求和,在继续学习时抬高适当性且不损失覆盖率。Protocol-level Rubrics(ProRubric) 研究的是:没有可验证器时,把 rubric 判分合成奖励的方式会不会让优化变成分数更高、回答更差。
用持续更新的探针做微调,降低有害性并保持线性可监测性
Published Sep 30
Evaluated models Mistral 7B Instruct v0.1 , Llama 3 8B Instruct Abliterated , Qwen3-14B Abstract 持续重拟合的探针可降低有害性、提高诚实性,训练后概念仍能被线性探针读出。Probe-guided fine-tuning 把探针分数当作唯一对齐信号,用来推动 harmlessness 与 honesty,并检查训练后这些概念是否仍线性可读。
分析隐蔽推理在思维链监控下的信息足迹下界与不可读性
Published Sep 29
Evaluated models Qwen2.5-7B-Instruct , Llama-3.1-8B-Instruct , Qwen 3.5 397B-A17B and others12 Abstract 论文从理论与实证探明了思维链监控的边界:复杂隐蔽计算必然泄露信息足迹且需承担长度税,但该足迹可被在线加密而不可读。
提出 STAR-GRPO,用成对可靠性约束抑制奖励作弊
Published Sep 29
Evaluated models Llama-3.2-1B-Instruct , Qwen3-4B Abstract STAR-GRPO 用成对可靠性约束 GRPO 优势,在两种奖励黑客设定中限制无支撑分数。STAR-GRPO 是一种可靠性优先的组相对优势估计:同一 rollout 的两个评分决定更新强度,任务奖励仍由预指定质量路径选择。
Proposes TIPS, inducing generative process supervision with only outcome rewards
Published Sep 29
Evaluated models Qwen3-4B-Thinking-2507 , Qwen3-4B-Instruct-2507 , Qwen3-8B-Thinking and others8 Abstract Outcome-supervised GRPO can induce step verification.TIPS trains a generative reward model using reinforcement learning with only outcome correctness, allowing step-level verification to appear in the chain of thought.