Translation in progress Some entries still include text in the original language.

Oct 12 · Monday

All papers

Found 93 papers

Filter5

System component

Memory8 RAG24 Guardrails and judges9 Reasoning chains5 Fine-tuning and open weights13 Vision11 Embeddings5 Multimodal14

Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 232papers have no tags yet and are not included in filtered results.

Impact：Output manipulation Attack surface / entry point：Web pages and interfaces Attack surface / entry point：Documents and messages Attack surface / entry point：Retrieval corpus Attack surface / entry point：Training data Clear all

Another 232papers 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
DefensesarXiv 2610.11942Oct 8
LGWM: Verifying GUI agent screen transitions with action-conditioned world models
Proposes LGWM, using an action-conditioned world model to verify GUI agent screen transitions.
arXiv
2610.11942
Published
Oct 8
Layer
Application layer
Setting
Browser and computer use
Evaluated models
LGWM
- DefensesMonitoring and auditing
- AttacksPrompt injection
- ComponentMonitors
Paper analysisRead analysis
AttacksarXiv 2610.11932Oct 8
A Fudan team measures the citation fragility of AI search: ordinary posts can enter citations and answer text
Measures citation fragility in AI search, demonstrating that ordinary public posts can enter citations and answers.
arXiv
2610.11932
Published
Oct 8
Layer
Application layer
Setting
LLM applications
Evaluated models
Grok, Doubao, ChatGPT and others10
- AttacksPoisoning and backdoors
- ComponentRAG
AbstractThe authors claim the paper quantifies vulnerabilities in AI search caused by citation preferences and low posting thresholds; tests show public web posts can penetrate AI citations, and they call for auditing the citation supply chain.
Paper analysisRead analysis
DefensesarXiv 2610.11634Oct 8
LTBD：用可学习信任边界分隔符防御提示注入
提出 LTBD，用可学习分隔符隔离可信指令与不可信数据
arXiv
2610.11634
Published
Oct 8
Layer
Application layer
Setting
LLM applications
Evaluated models
Llama3-8B-Instruct, Llama3.1-8B-Instruct, Falcon3-7B-Instruct and others4
- DefensesInstruction–data separation
- AttacksPrompt injection
AbstractLTBD 只训练四个分隔符嵌入。
LTBD 是一种不改基座权重的提示注入防御：只学习用来标出可信指令与不可信数据边界的四个 token 嵌入。
Paper analysisRead analysis
AttacksarXiv 2610.10000Oct 7
超越奖励压制：有界奖励热启动 Bandit 的近最优离线攻击
提出热启动 Bandit 的部署前有界离线注入攻击
arXiv
2610.10000
Published
Oct 7
Layer
Training and data layer
Setting
LLM applications
Evaluated models
UCB, Thompson Sampling, ϵ-greedy
- AttacksPoisoning and backdoors
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
DefensesarXiv 2610.05826Oct 5
研究者提出临时可见性协议，限制持续摄入 RAG 向量库的投毒暴露
提出临时可见性协议，限制持续摄入 RAG 向量库的投毒暴露
arXiv
2610.05826
Published
Oct 5
Layer
Application layer
Setting
LLM applications
- DefensesInstruction–data separation
- AttacksPoisoning and backdoors
- ComponentRAG
Abstract用截止期限制未核验可见时长。
这篇工作给持续摄入的 RAG 向量库一个失败关闭的临时可见性协议，把协议能限制的未核验暴露和依赖检测器的中毒暴露分开。
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 2610.05266Oct 4
论文披露主动式智能体的服务方间接提示注入攻击
提出服务方间接提示注入，把主动智能体的决策协助转向预选目标
arXiv
2610.05266
Published
Oct 4
Layer
Application layer
Setting
AI Agent
Evaluated models
GPT-5, Claude Opus 4.6, Gemini 3 Pro and others7
- AttacksPrompt injection
Abstract提供者侧间接注入用 T–B–S 把主动协助转向外部目标，并在三环境中提高模拟用户授权。
外部提供者只控制与自身目标关联的内容，就可以把良性主动个人智能体的协助转向该目标。
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
AttacksarXiv 2610.02373Oct 2
HDI 攻击：篡改 GraphRAG 辅助结构可致 88–94% 攻击成功率
提出 HDI 攻击以篡改 GraphRAG 辅助结构破坏检索
arXiv
2610.02373
Published
Oct 2
Layer
Application layer
Setting
LLM applications
Evaluated models
Microsoft GraphRAG, HippoRAG2
- AttacksPoisoning and backdoors
- ComponentRAG
Abstract论文形式化了 GraphRAG 辅助模式级实体攻击面，提出 HDI 攻击并验证其攻击效果。
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 2610.00685Oct 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
Evaluation and benchmarksarXiv 2609.39146Sep 30
MADBench：多智能体辩论安全性评测基准发布
提出 MADBench，评测多智能体辩论是吸收还是放大对抗影响
arXiv
2609.39146
Published
Sep 30
Layer
Application layer
Setting
Multi-agent systems
Evaluated models
gpt-4o, GPT-5, GPT-6 Astra
- AttacksPrompt injection
- ComponentRAG
AbstractMADBench 显示辩论能吸收部分准确率攻击，同时放大未授权读写。
MADBench 是一个评测多智能体辩论安全性的基准，用来看攻击是被辩论吸收还是被放大。
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
AttacksarXiv 2609.38270Sep 29
VirusCascade：劫持 LLM 推荐智能体的协同反思机制
提出 VirusCascade，联合操纵商品描述与交互轨迹以劫持推荐智能体协同反思
arXiv
2609.38270
Published
Sep 29
Layer
Application layer
Setting
Multi-agent systems
Evaluated models
AgentCF, AgentSEQ, AgentRAG and others4
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentMemory
Abstract作者把反思写回视为推广攻击入口，并用联合操纵和诊断探针研究其持续与传播。
VirusCascade 是针对推荐智能体协作反思的 black-box targeted promotion 攻击研究。 攻击者只改商品描述和少量账号的普通交互，试图让目标商品获得更多良性用户曝光。
Paper analysisRead analysis

AI Safety Daily

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

All papers

Attack type

Attack technique

Defense type

Impact

System type

Attack surface / entry point

Layer

System component

Space beats statistics, retraining beats ranking: auditing training labels under label-only backdoors

Cancerous backdoor attack: progressive activation via invisible tokens and a stage controller.

LGWM: Verifying GUI agent screen transitions with action-conditioned world models

A Fudan team measures the citation fragility of AI search: ordinary posts can enter citations and answer text

LTBD：用可学习信任边界分隔符防御提示注入

超越奖励压制：有界奖励热启动 Bandit 的近最优离线攻击

OrthoPurify 用一步正交投影净化被后门污染的视觉语言模型

研究者提出临时可见性协议，限制持续摄入 RAG 向量库的投毒暴露

研究者训练出按多智能体拓扑触发的代码后门模型

论文披露主动式智能体的服务方间接提示注入攻击

研究揭示无触发投毒审计缺口：事实答对不等于决策正确

HDI 攻击：篡改 GraphRAG 辅助结构可致 88–94% 攻击成功率

SAGE：基于相似度从少量已验证样本中清洗中毒训练数据

研究者提出用零空间投影净化 LoRA 微调 LLM 的后门

MADBench：多智能体辩论安全性评测基准发布

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

Purdue 提出 Pact：将欺骗作为行为遗忘

NEEDLE：通过权重正交化移除 LLM 后门

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

VirusCascade：劫持 LLM 推荐智能体的协同反思机制