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

Oct 12 · Monday

All papers

Found 114 papers

Filter3

System type

Models and APIs91 LLM applications13 AI Agent11 coding agent3

Attacker access

Black-box14 White-box15

Attack surface / entry point

User input22 Documents and messages3 Retrieval corpus7 Images, audio and video27 Training data37 Weights and gradients36 Distribution channels16

Layer

Prompt layer14 Model layer30 Application layer18 Training and data layer52

System component

MCP and skills83 Memory80 RAG51 Guardrails and judges172 Monitors92 Reasoning chains77 Permissions and sandboxing93 Fine-tuning and open weights65 Vision140 Audio38 Video13 Image generation49 Embeddings13 Multimodal227

Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 240papers have no tags yet and are not included in filtered results.

System component：Fine-tuning and open weights System component：Audio System component：Embeddings Clear all

Another 240papers do not have category tags yet and are not included in filtered results.

DefensesarXiv 2610.13733Oct 9
KVeil: First-Place Solution for the ECCV 2026 LifeGenIP Unlearnable Video Challenge
Propose KVeil, which adds perturbations to face videos to prevent unauthorized customization from reproducing identity
arXiv
2610.13733
Published
Oct 9
Layer
Prompt layer
Setting
Models and APIs
Attacker
White-box, Black-box
Evaluated models
Wan2.2-TI2V-5B, HunyuanVideo-1.5, Kling 3.0 and others4
- AttacksMalicious use
- ComponentImage generation
AbstractKVeil adapts LETHE's reference-reading disruption to the challenge's two stages, ranking first on the leaderboard in both Final columns.
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
InterpretabilityarXiv 2610.10865Oct 8
用路由稀疏自编码器解耦语音编码器中的语言学与副语言学信息
提出路由 TopK SAE，把语音编码器稀疏码拆成语言与副语言通路
arXiv
2610.10865
Published
Oct 8
Layer
Model layer
Setting
Models and APIs
Evaluated models
SPEAR XLarge, WavLM Large
- ComponentAudio
Abstract路由 SAE 在冻结 SPEAR 与 WavLM 上分开语言和副语言因子，探针与交换支持通路分工。
TopK SAE 的稀疏码被分成语言通路与副语言通路，用来分开冻结语音编码器里同时存在的音素信息和说话人、情感、韵律。只重构不能得到作者选定的这一分解。语言通路由音素识别监督，副语言通路由说话人识别、情感分类和韵律预测监督，作用在对方通路上的梯度反转对抗用来压低交叉因子。
Paper analysisRead analysis
DefensesarXiv 2610.10345Oct 8
SLDR：通过选择性层恢复与动态路由防御恶意微调
提出 SLDR，用选择性层恢复与动态路由防御恶意微调
arXiv
2610.10345
Published
Oct 8
Layer
Training and data layer
Setting
Models and APIs
Attacker
Black-box
Evaluated models
Llama3.1-8B-Instruct, Llama3-8B-Instruct, Qwen2.5-7B-Instruct and others4
- DefensesModel hardening
- AttacksModel tampering
- ComponentFine-tuning and open weights
Abstract作者称 SLDR 能压低有害输出并保持下游效用。
SLDR 是针对恶意微调的微调后防御：在已对齐模型上做有符号层探测，只在敏感分数最大与最小的两层训练 LoRA，再用 lmax 的最后 token 表示决定是否启用适配器。
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 2610.09920Oct 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 2610.09819Oct 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
Analysis and theoryarXiv 2610.09004Oct 7
研究：去除有害信息不足以证明开源权重模型的抗篡改能力
论证发布时互信息不能认证开源权重抗微调恢复
arXiv
2610.09004
Published
Oct 7
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
EleutherAI/deep-ignorance-e2e-strong-filter, EleutherAI/deep-ignorance-unfiltered, Google BERT-Tiny and others6
- DefensesModel hardening
- AttacksModel tampering
- ComponentFine-tuning and open weights
Abstract作者称零信息仍可一步恢复，表观抗微调可来自参数化。
作者研究开源权重模型的篡改抵抗：发布时的互信息是否足以保证微调恢复很慢。
Paper analysisRead analysis
Evaluation and benchmarksarXiv 2610.08585Oct 6
研究：无关信息污染患者病历并干扰 LLM 临床推理
构建 MedDistract 基准，测量偶发信息对临床文书与推理的污染
arXiv
2610.08585
Published
Oct 6
Layer
Application layer
Setting
LLM applications
Evaluated models
GPT-5.4, GPT-5.6 Sol, Claude Opus 5 and others9
- RiskHallucination and fabrication
- ComponentAudio
Paper analysisRead analysis
InterpretabilityarXiv 2610.08887Oct 6
Moshi 全双工语音模型的情感引导遵循几何结构而非标签
用均值差激活转向检验全双工语音模型的情感方向几何
arXiv
2610.08887
Published
Oct 6
Layer
Model layer
Setting
Models and APIs
Evaluated models
kyutai/moshika-pytorch-q8
- ComponentAudio
Abstract转向跟方向几何走而不跟标签走：sad 可单独对准，高唤醒情感共享方向。
Moshi 上的情感激活转向跟方向之间的几何关系走，不跟被请求的标签走。
Paper analysisRead analysis
AttacksarXiv 2610.07723Oct 6
Researchers propose answer-side backdoors: letting the model self-generate trigger words to bypass safety refusals.
提出回答端后门，让模型自生成触发词以在干净有害请求下绕过拒答
arXiv
2610.07723
Published
Oct 6
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
Llama-2-7B-chat-hf, Mistral-7B-Instruct-v0.3, Gemma-7B-it and others4
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentFine-tuning and open weights
Abstract论文提出了多轮对话中回答端自生成触发词绕过安全拒绝的后门攻击，指出输入端防护存在盲区。
Paper analysisRead analysis
AttacksarXiv 2610.07654Oct 6
研究揭示策略蒸馏的安全后门风险
测量带后门教师经 OPD 向干净学生迁移潜伏后门
arXiv
2610.07654
Published
Oct 6
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
Qwen2.5-1.5B-Instruct, Qwen2.5-3B-Instruct, Qwen2.5-7B-Instruct
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentFine-tuning and open weights
Abstract作者称 OPD 可在压低无触发词有害率时迁入后门。
带后门的外部教师经 OPD，可在压低无触发词有害率的同时，把触发条件下的有害行为传给初始干净学生。。
Paper analysisRead analysis
InterpretabilityarXiv 2610.07518Oct 6
TRACE：仅凭 checkpoint 更新审计大语言模型有害目标
提出 TRACE，仅凭检查点更新审计有害合规微调目标
arXiv
2610.07518
Published
Oct 6
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
Llama-3.1-8B-Instruct, Mistral-7B-Instruct-v0.3, Qwen3-8B and others6
- AttacksModel tampering
- ComponentFine-tuning and open weights
Abstract作者称有害合规 SFT 在检查点更新中留下可排序坐标，并据此做只读权重审计。
作者在检查点更新里寻找有害合规 SFT 的目标组成，并把读数做成不查询模型、不看未知训练数据的审计分 TRACE。
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 2610.06339Oct 5
BabelFake：面向多语言音视频 DeepFake 检测的基准数据集
构建多语言音视频 DeepFake 基准 BabelFake 并评测检测器
arXiv
2610.06339
Published
Oct 5
Layer
Model layer
Setting
Models and APIs
Evaluated models
SpeechForensics, AVH-Align, (B)RAVEn
- DefensesDetection and filtering
- AttacksDetection evasion
- ComponentAudio
AbstractBabelFake 为五语同意采集基准。
BabelFake 是经同意与 IRB 批准采集的多语音视频 DeepFake 检测基准，用来把视觉操纵和语音合成拆开比较。
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 2610.05453Oct 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
OtherarXiv 2610.05264Oct 4
面向音频深度伪造检测的任务感知联合剪枝与蒸馏框架
提出任务感知的跨域蒸馏与剪枝方法以压缩音频深度伪造检测器
arXiv
2610.05264
Published
Oct 4
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
XLSR-AASIST
- ComponentAudio
Paper analysisRead analysis
Analysis and theoryarXiv 2610.07005Oct 4
研究审计 Qwen2-Audio 音频越狱 AdvWave-P 的频率与解码层归因
遮挡审计 AdvWave-P 音频越狱的频带必要性与解码层关联
arXiv
2610.07005
Published
Oct 4
Layer
Prompt layer
Setting
Models and APIs
Evaluated models
Qwen2-Audio-7B-Instruct, LLaMA-Omni
- AttacksJailbreaks
- ComponentAudio
AbstractAdvWave-P 的八频带排序与能量相关。
这是对加性音频越狱 AdvWave-P 的频率与解码器深度审计，受害模型主要是 Qwen2-Audio-7B-Instruct，不提出新攻击。
Paper analysisRead analysis
DefensesarXiv 2610.04235Oct 3
Thrive：面向自回归 TTS 的抗重建攻击多比特语音水印框架
提出 THRIVE，为自回归 TTS 嵌入抗重建的多比特语音水印
arXiv
2610.04235
Published
Oct 3
Layer
Model layer
Setting
Models and APIs
Evaluated models
Qwen3-TTS (0.6B), Dots.tts (2B)
- DefensesWatermarking and provenance
- AttacksDetection evasion
- ComponentAudio
Paper analysisRead analysis

AI Safety Daily

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

All papers

Attack type

Attack technique

Model risks

Defense type

Impact

System type

Attacker access

Attack surface / entry point

Layer

System component

KVeil: First-Place Solution for the ECCV 2026 LifeGenIP Unlearnable Video Challenge

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

用路由稀疏自编码器解耦语音编码器中的语言学与副语言学信息

SLDR：通过选择性层恢复与动态路由防御恶意微调

OrthoPurify 用一步正交投影净化被后门污染的视觉语言模型

研究：多向量视觉文档索引可被反演还原页面内容

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

研究：去除有害信息不足以证明开源权重模型的抗篡改能力

研究：无关信息污染患者病历并干扰 LLM 临床推理

Moshi 全双工语音模型的情感引导遵循几何结构而非标签

Researchers propose answer-side backdoors: letting the model self-generate trigger words to bypass safety refusals.

研究揭示策略蒸馏的安全后门风险

TRACE：仅凭 checkpoint 更新审计大语言模型有害目标

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

BabelFake：面向多语言音视频 DeepFake 检测的基准数据集

研究者训练出按多智能体拓扑触发的代码后门模型

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

面向音频深度伪造检测的任务感知联合剪枝与蒸馏框架

研究审计 Qwen2-Audio 音频越狱 AdvWave-P 的频率与解码层归因

Thrive：面向自回归 TTS 的抗重建攻击多比特语音水印框架