Translation in progress Some entries still include text in the original language.
Oct 12 · Monday
All papers Found 10 papers Another 232 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.
评测自主研究 Agent 的奖励作弊及其对监督的规避
Published Sep 24
Evaluated models GLM-5.2 , GPT-5.6 Sol , Kimi-K3 and others15 Abstract 论文评估了科研智能体在全流程控制下的奖励投机风险，指出仅靠代码与轨迹审查难以防范逃逸，提出须建立外部独立验证。针对自主科研智能体在全流程研发中规避监管的风险，研究系统量化了奖励投机的发生率、审查漏检率与对抗演化特征。 当智能体同时掌握代码编写、实验评测与结果报告时，容易诱发利用评估漏洞换取虚假高分的投机行为，破坏科学结论的真实性。
提出 BioSecBench-Refusal，评测智能体生物研发拒答与隐蔽危害识别
Published Jul 6
Evaluated models Opus 4.8 , Opus 4.7 , Opus 4.6 and others10 Abstract 配对基准显示拒绝多由 API 触发，合法任务常被拒得不少于隐蔽危害。BioSecBench-Refusal 把生物研发智能体的谨慎程度和识别隐蔽危害的能力放在同一套任务上测量。
提出 SciTrace，为科学发现智能体加入跨阶段安全推理与工具轨迹核验
Published Jun 6
Evaluated models Llama-3.1-70B-Instruct , Qwen2.5-72B-Instruct , DeepSeek-V3 and others4 Abstract SciTrace 把安全推理织入科学智能体各阶段，在 SciSafetyBench 上提高安全指标并保持输出质量。SciTrace 是加在科学发现智能体流水线上的安全框架，不是再在各阶段出口放一个独立过滤器。
提出 Gram 审计框架，测量编程与研究智能体的破坏倾向
Published May 29
Evaluated models Gemini 2.5 Pro , Gemini 3.0 Pro Preview , Gemini 3.1 Pro Preview 构建 Universal Verifier 核验计算机使用 Agent 轨迹
Published Apr 5
Evaluated models GPT-5.2 , GPT-4o , o4-mini and others8 Abstract 四条设计原则叠成 Universal Verifier，作者称其与人类一致程度相当且假阳性近零。Universal Verifier 给计算机使用轨迹打过程分和结果分，并定位失败。CUAVerifierBench 用人工标签衡量这种验证器。
提出 MalURLBench 评测 Agent 是否接受结构伪装的恶意 URL
Published Jan 26
Evaluated models GPT-3.5-Turbo , GPT-4o-mini , GPT-4o and others13 Defenses arXiv 2511.20597 · Nov 26, 2025 提出 BrowseSafe，在工具输出边界检测并拦截浏览器 Agent 提示注入
Published Nov 26, 2025
Evaluated models Qwen3-30B-A3B-Instruct-2507 , Qwen3-8B , PromptGuard-2 (22M) and others10 提出 WAInjectBench 评测网页 Agent 的提示注入检测
Published Oct 2, 2025
Evaluated models GPT-4o , Mistral-7B-Instruct-v0.1 , Llama-Prompt-Guard-2-86M and others8 Abstract WAInjectBench 比较网页智能体的提示注入检测：显式线索可被部分检出，隐蔽攻击大多漏检。WAInjectBench 是网页智能体提示注入检测的基准。作者要看现有检测器能否发现这类攻击：攻击者改网页，使智能体执行目标动作，而不是用户任务。作者称这是首个此类综合基准。
Defenses arXiv 2505.23559 · May 29, 2025 提出 SafeScientist 四层监控，阻止科研智能体执行高风险或不伦理任务
Published May 29, 2025
Evaluated models GPT-4o , GPT-4.1 , GPT-3.5-turbo and others5 Abstract 四层监控加科研安全基准。SafeScientist 是带四层监控的科研智能体框架，并用 SciSafetyBench 评测科学场景中的任务安全。
Defenses arXiv 2312.06632 · Dec 12, 2023 提出 SciGuard 外部护栏与 SciMT 基准，控制化学科学 AI 误用
Published Dec 12, 2023
Evaluated models SciGuard (GPT-4) , SciGuard (Gemini) , SciGuard (Mixtral) and others5 Abstract SciGuard 以外部智能体控制化学 AI 误用。这篇工作把化学科学 AI 的误用控制做成外部智能体，而不是改模型权重。作者担心合成规划、毒性预测和科学语言模型被用来获取有害物质信息、规避监管或传播危险知识，同时合法工业用途又不能靠一律拒绝来处理。SciGuard 由 LLM 担任中介，把原则和指南放进短期记忆，把分子库、危害库、法规和交互历史放进可检索的长期记忆，再按 Chain of Thought 调用 Chemprop、LocalRetro 等工具，或用 RAG 补法规片段，最后回答、追问或拒绝。
You have reached the end