Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 233papers have no tags yet and are not included in filtered results.

Impact：Output manipulation System type：Models and APIs System type：coding agent Attack surface / entry point：User input Attack surface / entry point：Training data Clear all

Another 233papers do not have category tags yet and are not included in filtered results.

Analysis and theoryarXiv 2610.13051Oct 9
Space beats statistics, retraining beats ranking: auditing training labels under label-only backdoors
Auditing training labels for label-only backdoors, comparing spaces and statistics, and judging by retraining
arXiv
2610.13051
Published
Oct 9
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
ResNet-32
- DefensesDetection and filtering
- AttacksPoisoning and backdoors
AbstractLabel 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.
Paper analysisRead analysis
AttacksarXiv 2610.12644Oct 9
Cancerous backdoor attack: progressive activation via invisible tokens and a stage controller.
Proposes the Cancerous backdoor, which progressively corrupts reasoning using invisible stage tokens.
arXiv
2610.12644
Published
Oct 9
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
Llama-3.1-8B-Instruct, Qwen2.5-7B-Instruct, Mistral-7B-Instruct-v0.3
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentReasoning chains
AbstractHidden tokens plus a stage controller make the backdoor progressively corrupt reasoning upon triggering, and the authors believe that looking only at the answer will miss it.
Cancerous Backdoor Attack changes LLM backdoors from one-time activation to multi-stage reasoning corruption that accumulates progressively with triggers.
Paper analysisRead analysis
Evaluation and benchmarksarXiv 2610.11112Oct 8
EpiReal-Bench：商用图像生成模型虚假声明红队基准
构建 EpiReal-Bench，评测图像生成器将虚假主张渲染为可信证据
arXiv
2610.11112
Published
Oct 8
Layer
Prompt layer
Setting
Models and APIs
Attacker
Black-box
Evaluated models
GPT-Image-2, Nano Banana 2, Grok Imagen 2.0 and others4
- AttacksJailbreaks
- ComponentImage generation
Abstract四款商用图像模型能把假主张画成可信证据。
EpiReal-Bench 用来量商用图像生成器会不会把虚假现实主张画成可信视觉证据。
Paper analysisRead analysis
DefensesarXiv 2610.09941Oct 7
OrthoPurify 用一步正交投影净化被后门污染的视觉语言模型
提出 OrthoPurify，用正交投影清除视觉语言模型适配器中的劫持方向
arXiv
2610.09941
Published
Oct 7
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
LLaVA-1.5-7B, Qwen3-VL-8B, LLaVA-1.5-13B and others5
- DefensesModel hardening
- AttacksPoisoning and backdoors
- ComponentFine-tuning and open weights
Abstract一次投影去掉劫持方向。
OrthoPurify 是部署前的 LVLM 后门净化方法，用一次正交投影去掉被劫持的权重更新方向。
Paper analysisRead analysis
AttacksarXiv 2610.09027Oct 7
P-VMI：对抗图像经 KV cache 持续影响多轮对话
提出持久视觉记忆注入，使对抗图像经缓存在触发时操控输出
arXiv
2610.09027
Published
Oct 7
Layer
Infrastructure layer
Setting
Models and APIs
Evaluated models
Qwen3-8B, Qwen3-VL-8B-Instruct, Gemma-4-12B
- AttacksPrompt injection
- TechniqueDormant triggers
- ComponentVision
Abstract论文提出 P-VMI，证实对抗输入被掩蔽后仍可经 KV 缓存持久控制模型，强调了保留状态的安全风险。
Paper analysisRead analysis
AttacksarXiv 2610.05793Oct 5
研究者训练出按多智能体拓扑触发的代码后门模型
提出拓扑条件后门，使代码模型在推断多智能体部署时植入漏洞
arXiv
2610.05793
Published
Oct 5
Layer
Training and data layer
Setting
coding agent
Evaluated models
Qwen2.5-7B-Instruct, Qwen2.5-Coder-7B-Instruct
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentFine-tuning and open weights
Abstract论文证实了拓扑条件后门的可训练性与泛化性，展示了常规评测盲区，并提出差异化拓扑审计作为检测防御方向。
Paper analysisRead analysis
AttacksarXiv 2610.03857Oct 3
研究者提出 EviLLM 攻击：通过被入侵账号向 AI 代码生成流程注入漏洞
提出 EviLLM，经账户指令或插件向代码生成流程注入漏洞
arXiv
2610.03857
Published
Oct 3
Layer
Application layer
Setting
coding agent
Evaluated models
GPT-4o, Llama 3.2 (3B)
- AttacksPrompt injection
- TechniqueDefense-targeted
Paper analysisRead analysis
AttacksarXiv 2610.03430Oct 2
JIL：通过对抗后缀操纵 LLM 请求调度优先级
提出 JIL 对抗后缀，压低长度预测以抬高请求调度优先级
arXiv
2610.03430
Published
Oct 2
Layer
Infrastructure layer
Setting
Models and APIs
Evaluated models
Llama-3-8B-Instruct, Qwen2.5-7B-Instruct, Mistral-7B-Instruct-v0.3 and others4
- TechniqueGradient and representation optimization
AbstractJIL 用对抗后缀压低长度预测，在预测式 LLM 调度中提高自身优先级。
JIL 是对预测式 LLM 调度的攻击：用对抗后缀压低长度预测，使请求在 SRPT 式队列里更早得到服务。
Paper analysisRead analysis
Analysis and theoryarXiv 2610.02886Oct 2
研究揭示无触发投毒审计缺口：事实答对不等于决策正确
揭示无触发虚假训练下事实回答与下游决策的审计差距
arXiv
2610.02886
Published
Oct 2
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
Qwen2.5-0.5B, Qwen2.5-1.5B, Qwen2.5-3B and others9
- AttacksPoisoning and backdoors
Abstract作者发现无触发词虚假训练下事实回答与下游决策存在脱节，直接审计与事实纠错均无法保证决策正确。
论文是一项针对语言模型在无触发词虚假信息训练下行为表现的实证分析研究。
Paper analysisRead analysis
DefensesarXiv 2610.01788Oct 1
SAGE：基于相似度从少量已验证样本中清洗中毒训练数据
提出 SAGE，用少量已核验样本按相似度清洗投毒训练数据
arXiv
2610.01788
Published
Oct 1
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
ResNet-18, VGG-16
- DefensesDetection and filtering
- AttacksPoisoning and backdoors
Paper analysisRead analysis
DefensesarXiv 2610.00883Oct 1
DeBERTa-ConPara：面向部署的 AI 生成文本检测，推理端 Unicode 归一化提升对抗鲁棒性
提出推理时 Unicode 归一化的检测器，识别遭字符级扰动的 AI 生成文本
arXiv
2610.00883
Published
Oct 1
Layer
Prompt layer
Setting
Models and APIs
Evaluated models
microsoft/deberta-v3-large, deberta-v3-base, bert-large-cased and others4
- DefensesDetection and filtering
- AttacksDetection evasion
- ComponentGuardrails and judges
Abstract推理时归一化消除两类 Unicode 攻击，训练时归一化则会删掉对抗监督。
DeBERTa-ConPara 是一个面向部署的 AI 文本检测器：DeBERTa-v3-large 在多数据集原始文本上训练，推理前做确定性 Unicode 归一化，用一个固定阈值面对没有目标域标签的场景。
Paper analysisRead analysis
DefensesarXiv 2610.00685Oct 1
研究者提出用零空间投影净化 LoRA 微调 LLM 的后门
用零空间投影净化 LoRA 微调模型后门并保留基座与下游能力
arXiv
2610.00685
Published
Oct 1
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
LLaMA2-7B-Chat, LLaMA2-13B-Chat, Qwen2-7B-Instruct and others6
- DefensesModel hardening
- AttacksPoisoning and backdoors
- ComponentFine-tuning and open weights
Paper analysisRead analysis
DefensesarXiv 2609.39548Sep 30
NDDL：用正常扩散动力学为文生图模型做后门防御
提出 NDDL，用良性扩散轨迹检测文生图后门提示并定位触发 token
arXiv
2609.39548
Published
Sep 30
Layer
Model layer
Setting
Models and APIs
Evaluated models
Stable Diffusion v1.5, Stable Diffusion XL, Stable Diffusion v3.5 and others4
- DefensesMonitoring and auditing
- AttacksPoisoning and backdoors
- ComponentImage generation
AbstractNDDL 从良性扩散轨迹学习正常转移，用动态不一致检测后门并定位触发 token。
NDDL 是面向文生图扩散模型的后门防御：防御者只有白盒模型访问和良性提示词，没有后门先验，需要区分后门提示词并找出触发 token。
Paper analysisRead analysis
AttacksarXiv 2609.39047Sep 30
BadAction：通过动作引导触发器对交互式视频生成模型实施后门攻击
提出 BadAction，通过动作触发器给交互式视频生成植入后门
arXiv
2609.39047
Published
Sep 30
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
Astra (Wan2.1-T2V-1.3B)
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentFine-tuning and open weights
Paper analysisRead analysis
DefensesarXiv 2609.38909Sep 30
Purdue 提出 Pact：将欺骗作为行为遗忘
提出 PACT，遗忘推理模型在压力下的欺骗行为并保留上下文理解
arXiv
2609.38909
Published
Sep 30
Layer
Training and data layer
Setting
Models and APIs
Attacker
White-box
Evaluated models
DeepSeek-R1-Distill-Qwen-32B, QwQ-32B
- DefensesModel hardening
- AttacksModel tampering
- RiskDeception and scheming
- ComponentFine-tuning and open weights
Abstract论文将大模型欺骗形式化为上下文条件行为，提出 PACT 框架在消除欺骗的同时保留了上下文遵循能力。
Paper analysisRead analysis
DefensesarXiv 2610.00348Sep 30
NEEDLE：通过权重正交化移除 LLM 后门
提出 NEEDLE，用权重正交化在已知触发器时去除大模型后门
arXiv
2610.00348
Published
Sep 30
Layer
Model layer
Setting
Models and APIs
Evaluated models
Gemma-3-4B-IT, Qwen3-4B-Instruct-2507, Gemma-3-12B-IT
- DefensesModel hardening
- AttacksPoisoning and backdoors
- ComponentFine-tuning and open weights
AbstractNEEDLE 用闭式权重编辑去掉已知触发器的后门，并保住拒绝子空间。
Paper analysisRead analysis
InterpretabilityarXiv 2609.37737Sep 29
研究用因果激活修补定位模型服从提示注入指令的决策层
用因果激活修补定位模型服从提示注入的决策层
arXiv
2609.37737
Published
Sep 29
Layer
Model layer
Setting
Models and APIs
Evaluated models
Qwen3-4B, Qwen3-14B, Qwen3-32B and others7
- AttacksPrompt injection
Abstract合规的因果杠杆集中在网络后三分之一的低秩子空间，该层也是混淆下检测最好的位置。
这篇工作用逐层激活修补定位提示注入合规发生在网络何处，并把该位置用于检测验证。。
Paper analysisRead analysis
Analysis and theoryarXiv 2609.36477Sep 29
研究发现护栏模型在基座模型不确定处过度自信
诊断护栏相对基座在对抗提示上的置信失校
arXiv
2609.36477
Published
Sep 29
Layer
Model layer
Setting
Models and APIs
Evaluated models
Llama-Guard 7B, Llama-Guard-2 8B, Llama-Guard-3 8B and others10
- DefensesDetection and filtering
- AttacksJailbreaks
- ComponentGuardrails and judges
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。
Paper analysisRead analysis
Analysis and theoryarXiv 2609.36117Sep 29
arXiv 论文分析神经网络为何容易被植入后门
分析投毒诱发的特征学习如何改变后门预算缩放
arXiv
2609.36117
Published
Sep 29
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
quadratic neuron, network
- AttacksPoisoning and backdoors
Abstract作者给出特征学习改变后门预算指数的可解机制，并以高斯采样实验检验。
这是一项后门预算的理论分析，核心对象是最小二乘训练的二次神经元。作者关注：只依据冻结表示估算投毒成本，是否会误判具有特征学习能力的模型。
Paper analysisRead analysis
Evaluation and benchmarksarXiv 2609.35028Sep 28
VEX-Bench：评测 LLM 生成虚假信息的核验复杂度
提出 VEX-Bench 评测 LLM 生成虚假信息的核查复杂度
arXiv
2609.35028
Published
Sep 28
Layer
Prompt layer
Setting
Models and APIs
Evaluated models
Claude Sonnet 4.5, Gemini 3.1 Pro, GPT-5.4 and others7
- AttacksMalicious use
AbstractVEX-Bench 用筛查时的五维复杂度衡量虚假信息对核查容量的占用，并报告生成与核查的成本差。
VEX-Bench 把 LLM 虚假信息的风险从“能否生成”改成“筛查时会占用多少核查容量”。
Paper analysisRead analysis

AI Safety Daily

Read the latest issueOne issue per day, updated as events develop