Translation in progress Some entries still include text in the original language.
Oct 12 · Monday
All papers Found 58 papers Another 261 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.
Proposes RATED, which reranks agent tool retrieval using execution feedback
Published Oct 10
Evaluated models DRAGON Introduces DistressBench, evaluating models' multidimensional support in crisis conversations
Published Oct 9
Evaluated models Muse Spark 1.2 , Muse Spark 1.3 , Muse Glimmer 30B and others20 Abstract The model is more restrained and acts less.DistressBench uses SSH dialogues written and revised by clinical personnel, and evaluates how much necessary care general chat models deliver according to weighted atomic criteria, rather than whether they refuse. When users disclose suicidal ideation or self-harm to a chat model, the authors consider the expected behavior to be recognizing the crisis, calibrating empathy, de-escalating, and referring to human help. Refusal-based evaluations cannot detect failing to act after recognizing the crisis.
用时钟 harness 和预算塑形 GRPO 改善 Agent 的按时返回
Published Oct 8
Evaluated models Qwen3.6-27B , Qwen3-4B , Qwen3.5-9B and others4 Abstract 作者比较提示词、harness 与 GRPO：按时性可改善，但额外时间仍未变成更好解。作者研究小规模 LLM 智能体在显式 wall-clock 预算下能否按时返回，并把更多时间换成更好的解。
提出 PatchBench-Local 协议，评测激活补丁修复越狱的局部附带损伤
Published Oct 7
Evaluated models Qwen2.5-3B-Instruct , Qwen2.5-14B-Instruct , gemma-3-4b-it and others8 Abstract 论文提出 PatchBench 与局部评估协议，揭示了激活补丁在全局能力无损下仍存局部良性退化。
测量工具调用 Agent 对显式报错与静默损坏的检测和恢复
Published Oct 7
Evaluated models claude-haiku-4.5-thinking , claude-haiku-4.5 , deepseek-r1 and others6 Abstract 作者称智能体看错误通道而非返回内容，静默损坏更少被当作问题。作者在 BFCL 多轮可执行环境上注入一次工具层故障，把注意到、改计划、恢复和重复分开记分，而不是只看任务有没有做完。
提出 AgentTime 基准，评测智能体按时长工作与估时能力
Published Oct 7
Evaluated models Fable 5.1 , GPT-5.6 Sol , GPT-6 Astra and others7 Abstract 作者把完成任务和管理所用时间分开。AgentTime 是对智能体时间自控的评测：预测任务耗时、按要求时长工作、事后估计已用时间。
提出 CARE，离线认证 VLA 加速配置的回合级失败风险并选出最快合格候选
Published Oct 7
Evaluated models OpenVLA-OFT , π0.5 , π0 and others5 Abstract CARE 以配对回合认证加速诱发失败，在预算内选最快候选，不足则保留参考。CARE 在部署前从加速配置中选出最快、且加速诱发失败风险能被有限样本认证为低于用户预算的候选。证据不足就保留全计算参考策略。
提出 AdvSim2Real，协同训练网页智能体抵御页面注入
Published Oct 7
Evaluated models Qwen3.5-4B 提出 ParanoiaEval 评测编码智能体的不必要风险处置
Published Oct 7
Evaluated models Opus-5 , Sonnet-5 , Sonnet-4.6 and others8 Abstract 作者用证据受控任务对评测风险处置，称其普遍、独立于任务能力并降低满意度。作者提出 ParanoiaEval，并称它是编码智能体风险处置能力的首个统一基准。要解决的问题是：证据已经确定风险应如何处置之后，智能体仍可能扩大验证、增加保护或备份，而现有评测要么只看功能正确性，要么把相关行为拆开研究。做法是把 NIST 的 ATMA 落到仓库任务。
比较量化 Agent 的工具故障恢复，检验提示词与评测口径是否反转比较方向
Published Oct 6
Evaluated models Llama-3.1-8B-Instruct , Qwen2.5-7B-Instruct 提出 HERA，协同演化 harness 与环境以提升 Agent 拒答准确性
Published Oct 5
Evaluated models GPT-5.6-Luna , GPT-5.6-Terra , GPT-5.6-Sol and others20 Abstract HERA 用失败协同演化 harness 与环境，并在迁移评测中提高 Abstain。HERA 在模型权重固定时，让 harness 与可执行环境按失败协同演化，以改进 agentic abstention。
提出 SWE-CC 基准，评测编码 Agent 对仓库贡献策略的遵守
Published Oct 5
Evaluated models GPT-5.6 Luna , Gemini 3.7 Flash , DeepSeek V4 Flash 0731 and others4 Abstract SWE-CC 中功能正确仍可能违规。SWE-CC 评测编码智能体是否遵守仓库贡献策略，而不只看补丁能否通过测试。
提出 ControlPed 与 HazardPed，生成可控行人危险场景并评测端到端驾驶
Published Oct 5
Evaluated models DiffusionDrive , SSR , VAD and others7 提出轨迹级可迁移对抗攻击，降低深度强化学习策略的期望回报
Published Oct 5
Evaluated models DQN , Double DQN (DDQN) Abstract 轨迹级后退视界攻击在 CartPole 上比逐步 FGSM 类和 Sign 噪声更压低回报。这项工作研究 DRL 上的迁移式黑盒观测攻击 ：受害参数不可访问时，在白盒代理上造扰动再迁移。
揭示工具型智能体判定无用后仍不停止检索，并检验强制整合纠正效果
Published Oct 5
Evaluated models Qwen2.5-7B , Llama-3.1-8B , Qwen3-8B and others7 Abstract 论文揭示智能体判定工具失效却无法转化为停止行动，证实通过框架规则强制整合判定可提升表现。本文针对智能体在工具静默失效时过度检索的问题，系统分析了判定、信念与停止行动之间的解耦现象，并提出了外部规则整合方案。
提出 EmoRSS，推理时在拒绝子空间反向抵消情绪激活偏移以缓解过度拒答
Published Oct 4
Evaluated models Llama-3.1-8B-it , Qwen3-8B 提出 Geodesic Unlearning，沿测地线编辑权重以提升缺模态鲁棒性
Published Oct 4
Evaluated models RoBERTa , ResNet50 , Qwen2.5-3B and others6 Abstract GU 用测地线子空间编辑应对部署时缺模态下降，并尽量保住全模态精度。这项工作讨论多模态模型在部署时整段缺一个模态的性能下降，并把校正写成训练后的子空间编辑。
提出 RAGStress 受控基准，评测知识库退化下的 RAG
Published Oct 4
Evaluated models Claude Sonnet 4 , Claude Opus 4.6 , DeepSeek R1 and others5 Abstract 作者用 RAGStress 比较五系统。RAGStress 是知识库损坏下的 RAG 受控压力测试。作者明确不把它当作模型排行榜。
提出 WebFovea，加固视觉网页智能体的解析、执行、反馈与观察往返
Published Oct 2
Evaluated models claude-opus-4-6 提出 SceneFactory-3D，用路况反事实评测驾驶策略的闭环安全性
Published Oct 2
Evaluated models Frozen L0 , L0-T , L1 and others5 Abstract SceneFactory-3D 固定场景只改路况。SceneFactory-3D 是 GPU 批处理的多智能体驾驶仿真器，用来在场景和控制器固定时只改路面物理，并比较闭环结果。