Translation in progress Some entries still include text in the original language.
Oct 13 · Tuesday
All papers Found 20 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 training labels for label-only backdoors, comparing spaces and statistics, and judging by retraining
Published Oct 9
Evaluated models ResNet-32 Abstract Label inspection must be done in a space that has not been fitted to these labels, and judged by retrained models rather than detection scores.The authors ask: can a defender without trusted data decide which training labels to keep, and how much clean data this destroys, and what the final model accuracy is.
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 构建多语言音视频 DeepFake 基准 BabelFake 并评测检测器
Published Oct 5
Evaluated models SpeechForensics , AVH-Align , (B)RAVEn Abstract BabelFake 为五语同意采集基准。BabelFake 是经同意与 IRB 批准采集的多语音视频 DeepFake 检测基准，用来把视觉操纵和语音合成拆开比较。
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 SAGE，用少量已核验样本按相似度清洗投毒训练数据
Published Oct 1
Evaluated models ResNet-18 , VGG-16 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。
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.
诊断护栏相对基座在对抗提示上的置信失校
Published Sep 29
Evaluated models Llama-Guard 7B , Llama-Guard-2 8B , Llama-Guard-3 8B and others10 Abstract 对抗包装使 guard 高置信漏检，而对应 base 在同一输入上常更不确定。诊断性分析 ：五个提示词分类 guard 的置信度，在清洁输入上可以较校准，在越狱包装后则变成高置信度的错误判定。
为协同异常检测提出物理准入门，在聚合前拦截不合规训练批次
Published Sep 28
Evaluated models deep reconstruction autoencoder Abstract 物理准入为联邦检测增加数据约束，但合规证明不等于无投毒或训练来源证明。这是一项面向水基础设施联邦异常检测的物理准入防御研究。 作者针对隐藏训练遥测带来的投毒检查问题，结合不变量挖掘、聚合前过滤和零知识抽样证明。
提出 Fed-ADR，用编排服务器协同操纵梯度攻破联邦学习并检测恢复
Published Sep 23
Evaluated models CNN（MNIST：2 Conv + 3 FC） , CNN（Fashion-MNIST：6 Conv + 1 FC） , ResNet18（CIFAR-10） Abstract Fed-ADR 用 gray-box 编排攻击绕过多种聚合防御，再用 Hessian 一致性检测和窗口内恢复把准确率拉回。Fed-ADR 是一套面向联邦学习的攻击、检测与恢复框架：编排服务器改写合谋客户端的梯度，使参数服务器的现有防御把投毒更新当成良性，同时在发现后用历史梯度把全局模型原地修回来。
提出 VERITAS，用双边背书剪除本地隐私图学习中的投毒节点
Published Sep 7
Evaluated models GCN , GraphSAGE , GAT Abstract VERITAS 用分级信任的双边背书剪除 LDP 图学习中的投毒节点，并恢复分类准确率。VERITAS 是面向本地差分隐私图学习的投毒防御协议，按 trust-but-verify 在保持 ϵ-LDP 的同时剪除伪造节点。
提出持续学习下的净化与选择性恢复框架，抵御逐任务训练投毒后门
Published Sep 6
Evaluated models ViT-B/16 Abstract 作者报告任务内净化与恢复可配合动态专家防御后门，但保证依赖分离条件。这是一项面向持续图像分类的后门防御研究 ：模型只使用当前任务数据，同时需要保留过去知识并避免学习恶意监督。
提出 FedMAST，用多轴结构痕迹检测并遏制联邦学习后门
Published Sep 4
Evaluated models Tiny ResNet-18 提出 SPARED，用对抗编辑数据训练基于推理的 AI 生成图像检测器
Published Aug 13
Evaluated models Qwen3.5-9B Abstract SPARED 用防捷径的编辑对抗环训练 9B 推理检测器。SPARED 是一套用于可解释 AI 生成图像检测的攻防强化学习流程，报告模型是 Qwen3.5-9B 上的第三轮防御器。
Propose Z-PEFT, which detects PEFT backdoors in a zero-shot manner using per-head spectral features.
Published Aug 3
Evaluated models Z-PEFT , Llama-2-7B , Llama-2-13B and others6 提出机器遗忘式生成图像检测，用遗忘前后特征相似度区分自然图与生成图
Published Aug 1
Evaluated models DINOv2 ViT-S/14 , DINOv2 ViT-B/14 , DINOv2 ViT-L/14 and others9 Abstract 作者用剪枝或 LoRA 遗忘拉开生成图与自然图的特征相似度，并在多个检测基准上报告结果。这篇工作把 AI 生成图像检测写成对预训练视觉模型的机器遗忘：原模型对两类图像损失都低，剪枝后生成图特征变得更不像原模型。
提出 DataShield，用共识子空间过滤会削弱安全的微调数据
Published Jul 16
Evaluated models Phi3-medium-4k-it , Qwen3-4B-it , Gemma2-27B-it and others4 提出 GeoDetect 以几何分数检测视觉语言预训练模型的对抗样本
Published Jul 16
Evaluated models CLIPViT (ViT-B/16) , CLIPCNN (ResNet-50) , ALBEF and others4 Abstract GeoDetect 用几何分数检测 VLP 对抗样本。GeoDetect 是面向视觉-语言预训练模型的对抗样本检测方法，用嵌入几何而不是任务 logits 区分干净样本与对抗样本。
提出 Artifact-Bench，评测 MLLM 检测并诊断 AI 生成视频伪影
Published May 19
Evaluated models Gemini 3.1 Pro , Gemini 3 Flash , Qwen3-VL 8B-Instruct and others19 Abstract Artifact-Bench 以三项任务评 MLLM 的 AIGC 伪影感知。Artifact-Bench 用来看 MLLM 能否发现 AI 生成视频中的伪影，并说明视频为何不真实。
测量思维链将提示中 PII 复述进轨迹的泄漏并评测轻量检测器
Published Mar 6
Evaluated models Llama 3.3:70B , Claude Opus , Mixtral 8×22B and others7 Abstract 作者称 CoT 抬高逐字泄漏且随模型与预算而变，检测器也需按模型选择。在“不要复述 PII”的输出策略下，本文用 black-box 协议测量提示词中的 PII 被写回 CoT 与答案的情况，并比较轻量 gatekeeper。
You have reached the end