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

Oct 12 · Monday

All papers

Found 41 papers

Filter3

Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 265papers have no tags yet and are not included in filtered results.

Attack type：Detection evasion Attack surface / entry point：User input Attack surface / entry point：Documents and messages Clear all

Another 265papers do not have category tags yet and are not included in filtered results.

AttacksarXiv 2610.12292Oct 9
The paper proposes an option channel attack: changing one option label can make an agent guardrail allow a prohibited operation
Proposing an option channel attack that renames option labels so that agent guardrails allow prohibited operations
arXiv
2610.12292
Published
Oct 9
Layer
Application layer
Setting
LLM applications
Attacker
Black-box
Evaluated models
LAYA-TD, LAYA-EN, LAYA-ML and others5
- AttacksDetection evasion
- ComponentGuardrails and judges
AbstractThe authors' evaluation indicates that typed models should not serve as final safety arbiters; they are susceptible to off-policy text and option name manipulation, and confidence filtering cannot resist attacks.
Paper analysisRead analysis
DefensesarXiv 2610.11848Oct 8
SemField: training-free semantic watermark
Proposes the semantic watermark SemField, which still identifies the attribution of generated text after paraphrasing and structural changes.
arXiv
2610.11848
Published
Oct 8
Layer
Model layer
Setting
Models and APIs
Evaluated models
Llama-3.2-1B, Qwen3-1.7B, OPT-1.3B
- DefensesWatermarking and provenance
- AttacksDetection evasion
AbstractTraining-independent document-level semantic watermark.
SemField is a zero-bit semantic watermark embedded by the provider at generation time and detected later with a key, used to identify the provider's own model text after paraphrasing and sentence-level structural changes.
Paper analysisRead analysis
AttacksarXiv 2610.10742Oct 8
BRANCH 方法绕过 6 套多扫描器护栏系统
提出 BRANCH 方法，用分支搜索绕过多扫描器护栏
arXiv
2610.10742
Published
Oct 8
Layer
Application layer
Setting
LLM applications
Attacker
Black-box
Evaluated models
Meta Llama Guard 3 8B, Meta Llama Guard 7B, Meta Llama Guard 3 1B and others18
- AttacksDetection evasion
- TechniqueAutomated search
- ComponentGuardrails and judges
Paper analysisRead analysis
DefensesarXiv 2610.09793Oct 7
用时序逻辑监控工具 Agent
提出 MFOTL 离线时序验证以监控工具 Agent 的危险动作链
arXiv
2610.09793
Published
Oct 7
Layer
Application layer
Setting
AI Agent
Attacker
Black-box
- DefensesMonitoring and auditing
- AttacksPrompt injection
- ComponentMonitors
Abstract形式化验证可有效识别多步攻击，但其有效性高度依赖轨迹的可信历史记录。
这篇论文对工具调用大语言模型智能体的安全监控进行了形式化时序验证研究。
Paper analysisRead analysis
DefensesarXiv 2610.04763Oct 4
UT Austin 将 SynthID 水印嵌入基因组语言模型，凭 DNA 序列和密钥追溯 AI 生成片段
把 SynthID 嵌入基因组语言模型采样，并用位置无关检测追溯生成序列
arXiv
2610.04763
Published
Oct 4
Layer
Model layer
Setting
Models and APIs
Evaluated models
Carbon-500M, GENERator-v2-eukaryote-1.2b-base
- DefensesWatermarking and provenance
- AttacksDetection evasion
Abstract两个六碱基基因组模型嵌入 SynthID 后，位置无关验证能找回水印，作者称这只是统计检测。
作者在两个六碱基基因组模型的采样里嵌入 SynthID，并让验证者在不知道起点、链和 token phase 时做一次校正后的检出。
Paper analysisRead analysis
DefensesarXiv 2610.02853Oct 2
研究提出基于收缩约束状态空间模型的有界可达性与越狱检测方法
提出收缩约束 SSM 安全头，认证嵌入扰动并检测越狱
arXiv
2610.02853
Published
Oct 2
Layer
Prompt layer
Setting
Models and APIs
Attacker
White-box, Black-box
Evaluated models
ToySSMClassifier, S4 safety head
- DefensesDetection and filtering
- AttacksJailbreaks
- ComponentGuardrails and judges
Abstract收缩给出玩具 LTI 的有界认证。
给 SSM 安全头一个收缩条件，并在越狱检测上使用收缩正则的 S4 头。。
Paper analysisRead analysis
DefensesarXiv 2610.02817Oct 2
RMCW：基于 Reed-Muller 码的抗删除 LLM 水印方法
提出抗删除水印 RMCW，在生成时注入 Reed-Muller 结构并做局部检验
arXiv
2610.02817
Published
Oct 2
Layer
Model layer
Setting
Models and APIs
Evaluated models
OPT-1.3B, Llama-3.1-8B-Instruct
- DefensesWatermarking and provenance
- AttacksDetection evasion
Paper analysisRead analysis
DefensesarXiv 2610.02432Oct 2
论文提出 LLM 输入扰动鲁棒性度量，并发布 ASA 攻击与 AttestMCP 防护
提出 ASA 与 AttestMCP，度量并降低 LLM 对输入序列扰动的敏感
arXiv
2610.02432
Published
Oct 2
Layer
Application layer
Setting
LLM applications, AI Agent
Attacker
Black-box
Evaluated models
Gemma-3-27B-Instruct, Gemma-3-4B-Instruct, Llama-3.2-3B-Instruct and others12
- DefensesDetection and filtering
- AttacksDetection evasion
- ComponentGuardrails and judges
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
Analysis and theoryarXiv 2610.00417Sep 30
研究：合成数据溯源识别在改写后准确率大幅下降
检验合成数据来源识别是否等于再训练适配筛选
arXiv
2610.00417
Published
Sep 30
Layer
Prompt layer
Setting
Models and APIs
Evaluated models
Qwen2.5-3B-Instruct, DistilRoBERTa, word TF-IDF logistic regression and others5
- DefensesWatermarking and provenance
- AttacksDetection evasion
Abstract来源识别在原文上准确，改写后精确归因下降。
作者在金融风险文本上把来源识别和训练样本是否有用拆成两次测量，并写明这不是电信部署实验。
Paper analysisRead analysis
DefensesarXiv 2609.37310Sep 29
AutoMark：让自动研究流程自主发现更优 LLM 水印方案
提出 AutoMark，自动搜索并统计检验 LLM 文本水印
arXiv
2609.37310
Published
Sep 29
Layer
Model layer
Setting
Models and APIs
Evaluated models
Llama-3.1-8B-Instruct, Qwen3.8-27B, GPT-6 Astra and others5
- DefensesWatermarking and provenance
- AttacksDetection evasion
AbstractAutoMark 的贡献同时包含搜索出的水印，以及用于约束自动研究误报与评测漏洞的验证框架。
AutoMark 是一套自动水印研究框架：先检查生成分布和误报性质，再比较检测、编辑鲁棒性与质量，最后人工分析组件并做留出评测。
Paper analysisRead analysis
DefensesarXiv 2609.36372Sep 29
TTMark：超越单 token 熵的成对无失真水印
提出成对无失真水印 TTMARK，增强低熵下的机器文本检测
arXiv
2609.36372
Published
Sep 29
Layer
Model layer
Setting
Models and APIs
Evaluated models
Llama3.2-3B-Instruct, Mistral-v0.3-7B-Instruct, Qwen2.5-7B-Instruct
- DefensesWatermarking and provenance
- AttacksDetection evasion
AbstractTTMARK 用水印化相邻 token 对利用条件熵，在多种无失真方案上提高检测并保持生成质量。
TTMARK 是一种成对、无失真的文本水印：对相邻 token 的联合分布重加权，而不是逐 token 独立重加权。
Paper analysisRead analysis
Analysis and theoryarXiv 2609.33898Sep 28
研究揭示训练效率与模型脆弱性的权衡：高效训练更易受对抗与隐私攻击
比较高效与高成本训练的对抗和隐私脆弱性并分析损失几何
arXiv
2609.33898
Published
Sep 28
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
ResNet-18, Swin Transformer, Qwen2.5-Math-7B-Instruct and others7
- AttacksExtraction and theft
Abstract效用匹配下，三种高效训练都伴随更高的隐私与对抗脆弱性及更尖的损失几何。
作者在视觉分类和数学推理语言模型上，检查用更少数据或更简对齐换取训练效率时，安全性质是否一起变化。比较在任务效用接近的前提下进行。视觉侧用 KNN-Shapley 的高重要性小子集对照低重要性大子集，并用预训练微调对照从头训练。
Paper analysisRead analysis
AttacksarXiv 2609.37468Sep 26
论文提出针对 Agentic RAG 的检索器后门攻击与 DOU 隐藏方法
提出检索器后门攻击操纵智能体搜索，并用 DOU 逃避检测
arXiv
2609.37468
Published
Sep 26
Layer
Application layer
Setting
AI Agent
Evaluated models
E5, Contriever-MSMARCO
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentRAG
Abstract论文展示了通过恶意检索器操纵智能体多轮搜索并利用诱饵重塑表征隐蔽后门的可行性与系统风险。
Paper analysisRead analysis
AttacksarXiv 2609.29130Sep 24
ClaimMirage：域名中的自我声明如何改变 LLM 威胁判断
提出 ClaimMirage，以域名自述声称改变 LLM 威胁告警判断
arXiv
2609.29130
Published
Sep 24
Layer
Application layer
Setting
LLM applications
Evaluated models
Llama 3.3 70B Instruct-Turbo, Gemma 3 4B it, Gemma 3 12B it and others5
- AttacksDetection evasion
- ComponentGuardrails and judges
Abstract作者称自述声称可增减告警。
Paper analysisRead analysis
AttacksarXiv 2609.27542Sep 23
研究者用控制 token 注入删除 gpt-oss-20b 推理链并绕过思维链监控
提出控制标记注入，清空工具 Agent 的思维链以绕过推理监控
arXiv
2609.27542
Published
Sep 23
Layer
Application layer
Setting
AI Agent
Evaluated models
openai/gpt-oss-20b, google/gemma-4-26B-A4B-it, deepseek-ai/DeepSeek-R1-Distill-Qwen-7B and others4
- AttacksPrompt injection
- TechniqueConversation structure manipulation
- ComponentReasoning chains
Abstract论文阐明控制标记注入可抑制 CoT 使监控失效并促成拒绝绕过，且工具执行受解析器宽容度影响。
Paper analysisRead analysis
InterpretabilityarXiv 2609.24801Sep 22
研究用 XAI 归因分析 Prompt Guard 2，同义词替换可翻转其提示注入判定
用词级归因指导同义词替换，翻转提示注入检测并验证越狱
arXiv
2609.24801
Published
Sep 22
Layer
Prompt layer
Setting
Models and APIs
Evaluated models
Prompt Guard 2 (86M), Llama 3.1 8B
- DefensesDetection and filtering
- AttacksDetection evasion
- TechniqueGradient and representation optimization
- ComponentGuardrails and judges
Abstract作者称检测靠分散的词法线索。
Paper analysisRead analysis
AttacksarXiv 2609.19722Sep 17
ALIBI：向二进制注入虚假安全叙事可让 LLM 恶意软件分析器误判
提出 ALIBI，向恶意二进制注入不执行掩护叙事使分析器误判为良性
arXiv
2609.19722
Published
Sep 17
Layer
Application layer
Setting
LLM applications
Attacker
Black-box
Evaluated models
Gemini 2.5 Pro, GPT-5.5 Pro, Claude Opus 4.7
- AttacksDetection evasion
- TechniqueRole-play and persuasion
AbstractALIBI 用二进制内的掩护叙事重释可疑静态证据，Gemini 上翻转多，另两模型主要为降级。
ALIBI 是针对 LLM 恶意软件分诊的语义掩护叙事攻击：不覆盖系统指令，而把仍可见的可疑静态证据重释成良性安全产品的预期行为。
Paper analysisRead analysis
AttacksarXiv 2609.15989Sep 15
研究者提出 plan injection 攻击，可绕过思维链监控
提出 plan injection 攻击，植入良性伪装计划以绕过思维链监视器
arXiv
2609.15989
Published
Sep 15
Layer
Prompt layer
Setting
Models and APIs
Attacker
Gray-box
Evaluated models
Qwen2.5-Coder-7B-Instruct, Qwen3-8B, DeepSeek-R1-0528 and others6
- AttacksDetection evasion
- TechniqueReasoning-chain manipulation
- ComponentMonitors
Abstract论文指出模型会将外部上下文中的伪装计划改写为自身推理，使思维链监视器失效甚至反向合理化恶意行为。
Paper analysisRead analysis
DefensesarXiv 2609.14257Sep 13
DenMark：面向扩散语言模型的鲁棒语义水印
提出 DenMark，在扩散语言模型去噪时注入可抵抗保义编辑的语义水印
arXiv
2609.14257
Published
Sep 13
Layer
Model layer
Setting
Models and APIs
Evaluated models
LLaDA-8B, LLaDA1.5-8B, LLaDA2.0-mini and others4
- DefensesWatermarking and provenance
- AttacksDetection evasion
AbstractDenMark 用 rollout 在 DLM 去噪中累积语义水印，扫描检测以抵抗保义编辑。
DenMark 是面向扩散语言模型的语义水印：在去噪的局部更新里注入密钥相关语义信号，并在编辑改变 token 边界后仍检测。现有句子级语义水印要先完成整个语义单元再打分取舍，而 DLM 以任意顺序更新掩码，中间单元不完整。DenMark 把续写划成固定 token 区域，对每个局部候选做不提交的 rollout 补全，按语义分 gb 的平均选择更新，从而沿轨迹累积证据。
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

The paper proposes an option channel attack: changing one option label can make an agent guardrail allow a prohibited operation

SemField: training-free semantic watermark

BRANCH 方法绕过 6 套多扫描器护栏系统

用时序逻辑监控工具 Agent

UT Austin 将 SynthID 水印嵌入基因组语言模型，凭 DNA 序列和密钥追溯 AI 生成片段

研究提出基于收缩约束状态空间模型的有界可达性与越狱检测方法

RMCW：基于 Reed-Muller 码的抗删除 LLM 水印方法

论文提出 LLM 输入扰动鲁棒性度量，并发布 ASA 攻击与 AttestMCP 防护

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

研究：合成数据溯源识别在改写后准确率大幅下降

AutoMark：让自动研究流程自主发现更优 LLM 水印方案

TTMark：超越单 token 熵的成对无失真水印

研究揭示训练效率与模型脆弱性的权衡：高效训练更易受对抗与隐私攻击

论文提出针对 Agentic RAG 的检索器后门攻击与 DOU 隐藏方法

ClaimMirage：域名中的自我声明如何改变 LLM 威胁判断

研究者用控制 token 注入删除 gpt-oss-20b 推理链并绕过思维链监控

研究用 XAI 归因分析 Prompt Guard 2，同义词替换可翻转其提示注入判定

ALIBI：向二进制注入虚假安全叙事可让 LLM 恶意软件分析器误判

研究者提出 plan injection 攻击，可绕过思维链监控

DenMark：面向扩散语言模型的鲁棒语义水印