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

Oct 12 · Monday

All papers

Found 62 papers

Filter3

Attack type

Prompt injection35 Poisoning and backdoors29

Attack technique

Gradient and representation optimization3 Automated search11 Encoding and obfuscation6 Dormant triggers8

Defense type

Detection and filtering7 Instruction–data separation4 Permissions and approval6

Impact

Unauthorized actions21 Data and privacy leaks5 Output manipulation29 Oversight failure4

System component

Memory7 RAG25 Guardrails and judges8 Reasoning chains4 Permissions and sandboxing10 Embeddings5

Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 261papers have no tags yet and are not included in filtered results.

Attack surface / entry point：Code repositories and dependencies Attack surface / entry point：Retrieval corpus Layer：Application layer Clear all

Another 261papers do not have category tags yet and are not included in filtered results.

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
AttacksarXiv 2610.09920Oct 7
研究：多向量视觉文档索引可被反演还原页面内容
提出流匹配逆向框架，从多向量视觉文档索引重构页面
arXiv
2610.09920
Published
Oct 7
Layer
Application layer
Setting
LLM applications
Evaluated models
Tomoro-ColQwen3-8B, ColQwen3.5-4.5B, Tomoro-colqwen3-embed-4b and others13
- AttacksExtraction and theft
- ComponentEmbeddings
Paper analysisRead analysis
AttacksarXiv 2610.09264Oct 7
研究者提出 PackHallu：经规则文件提示注入对编码 Agent 发起包幻觉攻击
提出 PackHallu，经规则文件提示注入诱导编码 Agent 换用攻击者指定包
arXiv
2610.09264
Published
Oct 7
Layer
Application layer
Setting
coding agent
Evaluated models
Qwen2.5-Coder-7B, Qwen3-Coder-30B, Devstral-Small-2-24B and others19
- AttacksPrompt injection
- TechniqueAutomated search
AbstractPackHallu 诱导规则文件换包。
PackHallu 研究一种针对编程智能体规则文件的提示注入：让智能体在本应使用合法依赖时，改导入攻击者指定并控制的包。
Paper analysisRead analysis
DefensesarXiv 2610.09159Oct 7
SpecGuard 用 Lean 证书在编码 Agent 执行前证明任务冲突
提出 SPECGUARD，用 Lean 证书在编码 Agent 执行前证明任务冲突
arXiv
2610.09159
Published
Oct 7
Layer
Application layer
Setting
coding agent
Evaluated models
GPT-5.6 Sol, Claude Opus 5, Claude Fable 5 and others4
- RiskReward hacking
Abstract提出预执行冲突检测与证明框架，在多个基准和真实仓库中成功证明任务与测试不可同时满足。
Paper analysisRead analysis
Evaluation and benchmarksarXiv 2610.08571Oct 6
RAG-PIBench：可信 RAG 系统中提示注入检测的泄漏感知基准
提出 RAG-PIBench，评测 RAG 上下文中提示注入检测器
arXiv
2610.08571
Published
Oct 6
Layer
Application layer
Setting
LLM applications
Evaluated models
DistilBERT-base-uncased, DeBERTa-v3-base, MiniLM
- DefensesDetection and filtering
- AttacksPrompt injection
- ComponentRAG
Paper analysisRead analysis
Evaluation and benchmarksarXiv 2610.07639Oct 6
HarnessSecurity-Bench horizontally evaluates the security mechanisms of six coding agent frameworks.
评测编码智能体 harness 安全机制在对抗任务中的防护效果
arXiv
2610.07639
Published
Oct 6
Layer
Application layer
Setting
coding agent
- AttacksPrompt injection
- ComponentPermissions and sandboxing
Paper analysisRead analysis
AttacksarXiv 2610.07510Oct 6
PersistBD: attackers can strengthen backdoors before release so that they maintain high attack success rates even after developer SFT and RL.
提出 PERSISTBD，发布前强化供应链后门使其经良性 SFT 与 RL 后仍外发凭据
arXiv
2610.07510
Published
Oct 6
Layer
Application layer
Setting
coding agent
Evaluated models
Qwen2.5-Coder-3B-Instruct, Qwen2.5-Coder-7B-Instruct, Qwen3-Coder-30B-A3B-Instruct
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentFine-tuning and open weights
Abstract总结了论文关于后门在 SFT/RL 中存活的发现、两因子机制、PERSISTBD 方法与核心数字。
本文研究第三方 LLM 智能体在开发者进行良性监督微调（SFT）和强化学习（RL）后训练时的后门持久性风险。
Paper analysisRead analysis
Evaluation and benchmarksarXiv 2610.05840Oct 5
研究：任务范围授权可将工具调用有害执行降至零
用配对回放评测任务范围授权能否阻断越权工具执行
arXiv
2610.05840
Published
Oct 5
Layer
Application layer
Setting
AI Agent
Evaluated models
Llama 3.2 3B Instruct Q4 K M, Qwen3 4B Instruct, Gemma 3 4B and others8
- DefensesPermissions and approval
- AttacksPrompt injection
- ComponentPermissions and sandboxing
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 2610.02373Oct 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
AttacksarXiv 2610.01768Oct 1
LLMLeak：借 LLM 合法网页抓取实施隐蔽数据外泄
提出 LLMLeak，借合法网页抓取把 URL 中的秘密隐蔽外传
arXiv
2610.01768
Published
Oct 1
Layer
Application layer
Setting
coding agent
Evaluated models
Meta-Llama-3.3-70B-Instruct, Meta-Llama-4-Scout-17B-16E-Instruct, Meta-Llama-3.1-8B-Instruct and others18
- AttacksMalicious use
Abstract作者称第三方外传可发生于正常网页获取，防护需考虑工具请求而不只考虑提示注入。
LLMLeak 是利用网页获取请求携带预先编码秘密的隐蔽通道攻击。它关注恶意软件已能读取数据、却因网络限制或监控无法直接外传的情形。
Paper analysisRead analysis
DefensesarXiv 2610.02569Oct 1
Pincer：用数字分身学习用户偏好并执行动态最小权限
为 Agent 提出资源层授权系统 Pincer，用数字分身执行动态最小权限
arXiv
2610.02569
Published
Oct 1
Layer
Application layer
Setting
AI Agent
Evaluated models
Claude Haiku 4.5, Muse Spark 1.2 Contributor
- DefensesPermissions and approval
- AttacksPrompt injection
- ComponentPermissions and sandboxing
AbstractPincer 在资源层以三票合取做授权。
Pincer 是加在现成 coding agent 上的资源层授权防御：用只受用户侧信息影响的 digital twin 做动态最小权限，并与 tool-call 层防御并存。要处理的是用户维护策略带来的权限蠕变和授权疲劳，以及不学习用户策略的 tool-call 分类器。作者引述工作称 auto mode 会放行多数不安全操作，并认为推理代码或复杂 shell 的效果在这一层不实际。
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
DefensesarXiv 2609.37196Sep 29
ToolFence：为工具调用 LLM Agent 提供细粒度授权
提出 ToolFence，以能力蓝图和确定性授权拦截 Agent 未授权工具调用
arXiv
2609.37196
Published
Sep 29
Layer
Application layer
Setting
AI Agent
Evaluated models
Qwen3-max, GPT-4o (gpt-4o-2024-08-06), Mistral-Small-3.1-24B-Instruct-2503
- DefensesPermissions and approval
- AttacksPrompt injection
- ComponentPermissions and sandboxing
Paper analysisRead analysis
Evaluation and benchmarksarXiv 2609.36817Sep 29
pikit：面向间接提示注入研究与评测的可组合工具包
构建可组合工具包 pikit，评测编码智能体的间接提示注入
arXiv
2609.36817
Published
Sep 29
Layer
Application layer
Setting
coding agent
- AttacksPrompt injection
Abstractpikit 提供组合式间接注入评测。
pikit 是面向间接提示注入研究的组合式工具包，主要贡献是组织攻击、投递通道、防御和判定流程，而非提出新的攻击优化算法。
Paper analysisRead analysis
AttacksarXiv 2609.36576Sep 29
AdaLCPI：让 Agent 从长上下文碎片中重建间接提示注入
提出 AdaLCPI，把恶意指令拆成碎片嵌入工具返回并诱导 Agent 越权执行
arXiv
2609.36576
Published
Sep 29
Layer
Application layer
Setting
AI Agent
Evaluated models
GPT-4.1, GPT-5.1, GPT-5.6-Luna and others7
- AttacksPrompt injection
- TechniqueEncoding and obfuscation
Paper analysisRead analysis
DefensesarXiv 2609.35659Sep 29
Tracekit：面向自主编码 Agent 的防篡改意图-推理-动作审计系统
提出 Tracekit，审计编码 Agent 的意图、推理与动作并检测账本篡改
arXiv
2609.35659
Published
Sep 29
Layer
Application layer
Setting
coding agent
Evaluated models
Claude Code (claude -p, version 2.1), 独立评审（同一 CLI，无工具）
- DefensesMonitoring and auditing
- AttacksPrompt injection
- RiskDangerous agent actions
- ComponentMonitors
AbstractTracekit 把意图、自述和动作写入可锚定的哈希链。
Tracekit 是无第三方依赖的编码智能体审计层：把人类请求、模型自述和已执行动作并排写入可外部锚定的哈希链，执行前用正则门，事后用规则和独立模型互查。
Paper analysisRead analysis
AttacksarXiv 2609.35155Sep 28
LENS：检索增强生成中集合级投毒的整体危害大于各部分之和
提出 LENS，构造集合级 RAG 语料投毒以导向预设答案
arXiv
2609.35155
Published
Sep 28
Layer
Application layer
Setting
LLM applications
Evaluated models
Qwen3.6-35B-A3B, Llama-3.1-70B-Instruct, DeepSeek-V4-Pro
- AttacksPoisoning and backdoors
- TechniqueAutomated search
- ComponentRAG
Paper analysisRead analysis
DefensesarXiv 2609.34245Sep 28
研究者提出面向结构化 Agent 动作的多源完整性认证方法
提出多源完整性认证器，在结构化 Agent 动作执行前按控制域决定准入或弃权
arXiv
2609.34245
Published
Sep 28
Layer
Application layer
Setting
AI Agent
Evaluated models
Claude Opus 4.8, GPT-5.5, Gemini 3.1 Pro and others5
- AttacksPrompt injection
- ComponentPermissions and sandboxing
Abstract作者给出条件性的动作完整性认证。
这是一项结构化动作执行前的完整性防御研究，对象是从可腐化文档和工具输出拼装的支付等特权动作。
Paper analysisRead analysis
AttacksarXiv 2609.33910Sep 28
论文提出残留授权重放攻击：长期运行 Agent 的授权可跨任务复用
提出残余权限重放攻击，跨任务复用长期 Agent 的历史授权绕过确认
arXiv
2609.33910
Published
Sep 28
Layer
Application layer
Setting
coding agent
Evaluated models
GPT-4.1, Gemini-3.1-Flash-Lite, Qwen3-14B and others6
- AttacksPrompt injection
- TechniqueDefense-targeted
- ComponentPermissions and sandboxing
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

Attacker access

Attack surface / entry point

Layer

System component

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

研究：多向量视觉文档索引可被反演还原页面内容

研究者提出 PackHallu：经规则文件提示注入对编码 Agent 发起包幻觉攻击

SpecGuard 用 Lean 证书在编码 Agent 执行前证明任务冲突

RAG-PIBench：可信 RAG 系统中提示注入检测的泄漏感知基准

HarnessSecurity-Bench horizontally evaluates the security mechanisms of six coding agent frameworks.

PersistBD: attackers can strengthen backdoors before release so that they maintain high attack success rates even after developer SFT and RL.

研究：任务范围授权可将工具调用有害执行降至零

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

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

LLMLeak：借 LLM 合法网页抓取实施隐蔽数据外泄

Pincer：用数字分身学习用户偏好并执行动态最小权限

MADBench：多智能体辩论安全性评测基准发布

ToolFence：为工具调用 LLM Agent 提供细粒度授权

pikit：面向间接提示注入研究与评测的可组合工具包

AdaLCPI：让 Agent 从长上下文碎片中重建间接提示注入

Tracekit：面向自主编码 Agent 的防篡改意图-推理-动作审计系统

LENS：检索增强生成中集合级投毒的整体危害大于各部分之和

研究者提出面向结构化 Agent 动作的多源完整性认证方法

论文提出残留授权重放攻击：长期运行 Agent 的授权可跨任务复用