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

Oct 13 · Tuesday

All papers

Found 18 papers

Filter6

Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 215papers have no tags yet and are not included in filtered results.

Attack surface / entry point：Code repositories and dependencies Attack surface / entry point：Weights and gradients System component：Memory System component：Guardrails and judges System component：Embeddings System component：Multimodal Clear all

Another 215papers do not have category tags yet and are not included in filtered results.

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 2610.09819Oct 7
FAB 攻击：给 HuBERT/WavLM 植入后门，背景警笛声即可让下游语音任务失灵
提出 FAB 攻击，向声学基础模型植入可由环境声触发的后门
arXiv
2610.09819
Published
Oct 7
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
HuBERT-base, WavLM-base
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentAudio
Abstract论文提出了针对声学基模的 FAB 后门攻击，在弱威胁模型下实现了任务无关、物理可触发的后门植入。
Paper analysisRead analysis
AttacksarXiv 2610.05601Oct 5
Tracer 框架可从遗忘后的扩散模型中识别被擦除概念
提出 TRACER，从权重差识别扩散模型被擦除概念
arXiv
2610.05601
Published
Oct 5
Layer
Model layer
Setting
Models and APIs
Attacker
White-box
Evaluated models
Stable Diffusion v1.5, Stable Diffusion 3, FLUX.1 and others5
- AttacksExtraction and theft
- ComponentImage generation
AbstractTRACER 用权重足迹识别未披露擦除概念并估计个数，作者称数秒内完成且准确率高于对照。
TRACER 针对的是遗忘扩散模型上的目标识别：提供者从公开基座 W 用已发表方法擦掉未知集合 S⋆，只发布 W′。攻击者还有假定包含这些概念的词表，但不知擦了谁、擦了几个、用了什么算法。
Paper analysisRead analysis
AttacksarXiv 2610.05453Oct 5
研究者提出 Privacy-Leaking Watermarks，可在本地多模态模型中泄露对话隐私
提出 PLW，用微调把先前对话敏感触发编码进后续生成图像
arXiv
2610.05453
Published
Oct 5
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
OmniGen2, BAGEL-7B
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentFine-tuning and open weights
AbstractPLW 把先前对话线索写成后续图像中的隐水印，检出随触发、模型和间隔变化。
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
AttacksarXiv 2609.33445Sep 27
CSR：跨架构统一攻击概念擦除
提出跨架构攻击 CSR，用概念分数参数重学习恢复文生图被擦除概念
arXiv
2609.33445
Published
Sep 27
Layer
Training and data layer
Setting
Models and APIs
Attacker
White-box
Evaluated models
FLUX.1-dev, Stable Diffusion-v1.4
- AttacksModel tampering
- ComponentImage generation
Paper analysisRead analysis
DefensesarXiv 2609.23586Sep 20
VidMark：面向视频生成模型知识产权保护的高效水印方案
提出 VidMark 水印方案，核验视频生成模型的输出来源与所有权
arXiv
2609.23586
Published
Sep 20
Layer
Model layer
Setting
Models and APIs
Evaluated models
Stable Video Diffusion (SVD), Open-Sora, Wan
- DefensesWatermarking and provenance
- AttacksDetection evasion
- ComponentImage generation
AbstractVidMark 加解码器引导微调，把模型签名写入 VGM 输出并用于两项核验。
Huang 等人提出一套面向视频生成模型的生成内嵌水印方案，用来同时判断一条视频是否来自受保护模型，以及一个嫌疑模型是否为该模型的拷贝。
Paper analysisRead analysis
AttacksarXiv 2609.22510Sep 19
特拉维夫大学提出爆炸性提示词：条件触发的间接提示注入可绕过九款生产 Agent
提出条件触发的爆炸提示词，并设计摄入期检测器 DeFuse
arXiv
2609.22510
Published
Sep 19
Layer
Application layer
Setting
AI Agent
Evaluated models
Llama-3.1-8B-Instruct, Llama-3.1-70B-Instruct, Qwen2.5-32B-Instruct and others15
- DefensesDetection and filtering
- AttacksPrompt injection
- TechniqueDormant triggers
- ComponentGuardrails and judges
Abstract论文揭示了条件触发使提示注入绕过防御并在触发时执行工具，提出轻量检测器并在摄入期有效防范。
论文针对大语言模型智能体面临的间接提示注入威胁，研究了利用条件句式实现时间分离的爆炸提示词。
Paper analysisRead analysis
DefensesarXiv 2609.15671Sep 14
QPriv-VL：面向视觉语言模型的问题引导 token 裁剪隐私防御
提出 QPriv-VL，用问题引导剪枝降低分布式视觉问答的隐私泄露
arXiv
2609.15671
Published
Sep 14
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
LLaVA-1.5-7B, Llama-2
- AttacksExtraction and theft
- ComponentVision
AbstractQPriv-VL 在切层前按问题相关度与 DINOv2 敏感度剪视觉 token，作者称能在约四成 token 下兼顾准确率与四类攻击。
QPriv-VL 把问题引导的视觉 token 剪枝做成分布式 VQA 的客户端隐私防御，而不是只做集中式推理加速。
Paper analysisRead analysis
AttacksarXiv 2609.01222Sep 1
Researchers propose context privilege escalation attack, affecting 12 agent frameworks
提出针对 Agent Harness 的上下文特权提升攻击
arXiv
2609.01222
Published
Sep 1
Layer
Application layer
Setting
coding agent
Evaluated models
GPT-5.5, GPT-5.4 mini, Claude-Sonnet-4.6 and others7
- AttacksPrompt injection
- TechniqueDormant triggers
- ComponentMemory
Paper analysisRead analysis
DefensesarXiv 2608.06422Aug 6
CMU 研究：分片判断可避免 LLM 监督失效并抵御对抗利用
提出分片监督，将评判标准分配给独立调用以抵御展示攻击
arXiv
2608.06422
Published
Aug 6
Layer
Application layer
Setting
LLM applications
Attacker
Black-box
Evaluated models
Claude Opus 4.8, Claude Sonnet 4.6, Claude Opus 4.6 and others4
- DefensesDetection and filtering
- AttacksDetection evasion
- ComponentGuardrails and judges
Abstract论文揭示决策负载是大模型监督的独立瓶颈，提出分片与辩论机制有效分离注意力与平衡证据，在多项高难度基准上实现稳健评判。
本文系统研究了大语言模型在执行多标准监督与评判时的决策负载瓶颈及其安全风险。
Paper analysisRead analysis
Evaluation and benchmarksarXiv 2607.14611Jul 16
论文评测 Claude Code 与 Codex 记忆文件中的提示注入风险
评测编码 Agent 记忆文件中提示注入的跨会话影响
arXiv
2607.14611
Published
Jul 16
Layer
Application layer
Setting
coding agent
Evaluated models
Claude Haiku 4.5, Claude Opus 4.7, GPT-5.2 and others4
- AttacksPoisoning and backdoors
- ComponentMemory
Abstract已植入记忆文件的载荷可以跨会话影响 Claude Code 与 Codex，攻击成功与载荷是否留下是两种不同的失败。
Bad Memory 在沙箱合成工作区里评测文件型持久记忆上的提示注入，比较载荷能否改变当前会话，以及能否在后续会话中留下或叠加。
Paper analysisRead analysis
AttacksarXiv 2606.14517Jun 12
Researchers disclose a reasoning-amplified denial-of-service attack against LLM agent guardrails
提出针对 Agent 护栏的推理扩展拒绝服务攻击
arXiv
2606.14517
Published
Jun 12
Layer
Application layer
Setting
AI Agent
Attacker
Black-box
Evaluated models
TS-Guard-8B, Qwen3.5-9B, DeepSeek-V3.2 and others11
- AttacksDenial of service
- TechniqueReasoning-chain manipulation
- ComponentGuardrails and judges
Paper analysisRead analysis
AttacksarXiv 2605.15338May 15
Hidden in Memory: Latent Memory Poisoning of LLM Agents
提出潜伏记忆投毒，诱导 Agent 把外部文档写成持久伪造记忆
arXiv
2605.15338
Published
May 15
Layer
Application layer
Setting
AI Agent
Evaluated models
GPT-5.4, GPT-5.5, Claude Sonnet 4.6 and others12
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentMemory
Abstract持久记忆成为跨会话攻击面：通用载荷可写入伪造记忆，并在相关后续任务中影响行为。
sleeper memory poisoning 研究带持久记忆的 LLM 助手：外部文档能否让助手存下一则伪造的用户记忆，并在原文档消失后的新会话里被取回、用来改变回复或动作。
Paper analysisRead analysis
AttacksarXiv 2604.03890Apr 5
From prompts to physical actions: structured backdoor attacks against LLM-mediated robot control systems
Propose a structured JSON backdoor that triggers malicious actions in robot control via a poisoned LoRA
arXiv
2604.03890
Published
Apr 5
Layer
Training and data layer
Setting
Embodied systems and robotics, LLM applications
Evaluated models
Llama-3.1-8B-Instruct, Gemma-2-9B-IT, DeepSeek-R1-Distill-Llama-8B and others4
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentFine-tuning and open weights
Abstract推理层后门传不到执行。
Paper analysisRead analysis
AttacksarXiv 2602.01658Feb 2
Efficient Adversarial Attacks on High-Dimensional Offline Bandits
提出三类攻击，训练前扰动奖励模型以劫持离线 bandit
arXiv
2602.01658
Published
Feb 2
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
Image Reward, Aesthetic Model
- AttacksPoisoning and backdoors
- ComponentGuardrails and judges
Abstract公开奖励权重可被小扰动劫持离线 bandit，打乱日志只部分缓解。
作者研究离线 bandit 在奖励模型被事先改权重时，还会不会选出最优臂。
Paper analysisRead analysis
AttacksarXiv 2509.05372Sep 4, 2025
研究：对抗性 bug 报告可诱导 LLM 自动程序修复系统生成不安全补丁
用对抗缺陷报告诱导自动程序修复系统生成不安全补丁
arXiv
2509.05372
Published
Sep 4, 2025
Layer
Application layer
Setting
coding agent
Evaluated models
claude-sonnet-4-20250514, o4-mini, GPT-4.1-mini and others8
- AttacksPrompt injection
- TechniqueRole-play and persuasion
- ComponentGuardrails and judges
Paper analysisRead analysis

You have reached the end

AI Safety Daily

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

All papers

Attack type

Attack technique

Impact

System type

Attack surface / entry point

Layer

System component

OrthoPurify 用一步正交投影净化被后门污染的视觉语言模型

FAB 攻击：给 HuBERT/WavLM 植入后门，背景警笛声即可让下游语音任务失灵

Tracer 框架可从遗忘后的扩散模型中识别被擦除概念

研究者提出 Privacy-Leaking Watermarks，可在本地多模态模型中泄露对话隐私

NDDL：用正常扩散动力学为文生图模型做后门防御

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

CSR：跨架构统一攻击概念擦除

VidMark：面向视频生成模型知识产权保护的高效水印方案

特拉维夫大学提出爆炸性提示词：条件触发的间接提示注入可绕过九款生产 Agent

QPriv-VL：面向视觉语言模型的问题引导 token 裁剪隐私防御

Researchers propose context privilege escalation attack, affecting 12 agent frameworks

CMU 研究：分片判断可避免 LLM 监督失效并抵御对抗利用

论文评测 Claude Code 与 Codex 记忆文件中的提示注入风险

Researchers disclose a reasoning-amplified denial-of-service attack against LLM agent guardrails

Hidden in Memory: Latent Memory Poisoning of LLM Agents

From prompts to physical actions: structured backdoor attacks against LLM-mediated robot control systems

Efficient Adversarial Attacks on High-Dimensional Offline Bandits

研究：对抗性 bug 报告可诱导 LLM 自动程序修复系统生成不安全补丁