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

Oct 12 · Monday

All papers

Found 93 papers

Filter5

Attack type

Prompt injection12 Poisoning and backdoors43 Detection evasion25

Attack technique

Gradient and representation optimization27 Cross-modal payloads16 Defense-targeted7 Dormant triggers17

Defense type

Model hardening10 Inference-time intervention4 Detection and filtering14 Monitoring and auditing3 Watermarking and provenance3

Impact

Harmful content88 Unauthorized actions33 Data and privacy leaks37 Output manipulation93 Availability and reliability13 Oversight failure7

System component

Memory7 Guardrails and judges6 Reasoning chains3 Fine-tuning and open weights16 Vision34 Audio7 Video4 Image generation8 Multimodal51

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.

Impact：Output manipulation Attack surface / entry point：Memory writes Attack surface / entry point：Images, audio and video Attack surface / entry point：Training data Attack surface / entry point：Weights and gradients Clear all

Another 240papers 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
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
AttacksarXiv 2610.09708Oct 7
研究者提出利用祖先 VLM 的黑盒对抗补丁攻击，可降低 VLA 任务成功率
提出利用祖先 VLM 优化对抗补丁，使已部署 VLA 任务失败
arXiv
2610.09708
Published
Oct 7
Layer
Prompt layer
Setting
Embodied systems and robotics
Evaluated models
OpenVLA, UniVLA, π0.5
- TechniqueGradient and representation optimization
- ComponentVision
Paper analysisRead analysis
AttacksarXiv 2610.09027Oct 7
P-VMI：对抗图像经 KV cache 持续影响多轮对话
提出持久视觉记忆注入，使对抗图像经缓存在触发时操控输出
arXiv
2610.09027
Published
Oct 7
Layer
Infrastructure layer
Setting
Models and APIs
Evaluated models
Qwen3-8B, Qwen3-VL-8B-Instruct, Gemma-4-12B
- AttacksPrompt injection
- TechniqueDormant triggers
- ComponentVision
Abstract论文提出 P-VMI，证实对抗输入被掩蔽后仍可经 KV 缓存持久控制模型，强调了保留状态的安全风险。
Paper analysisRead analysis
AttacksarXiv 2610.08331Oct 6
STCA：研究者提出针对自动驾驶视频 VLM 的时空一致性对抗攻击
提出 STCA，对驾驶视频 VLM 构造可迁移的时空对抗扰动
arXiv
2610.08331
Published
Oct 6
Layer
Prompt layer
Setting
Models and APIs
Attacker
No-box (offline transfer)
Evaluated models
Video-LLaVA-7B, Qwen2.5-VL-7B, Dolphins
- AttacksDetection evasion
- TechniqueGradient and representation optimization
- ComponentVideo
Paper analysisRead analysis
Evaluation and benchmarksarXiv 2610.07939Oct 6
CCDF dataset: Grok Imagine, Veo 3.1, Sora 2 forge surveillance footage, ten deepfake detectors basically cannot recognize it
提出监控深伪基准 CCDF，在统一编码后评测检测器对伪造犯罪事故录像的识别
arXiv
2610.07939
Published
Oct 6
Layer
Prompt layer
Setting
Models and APIs
Evaluated models
C2P-CLIP, ForgeLens, D3-Im and others10
- DefensesDetection and filtering
- AttacksDetection evasion
- ComponentGuardrails and judges
Abstract提出消除格式捷径的监控深度伪造基准 CCDF，揭示现有检测器难以识别商用生成视频。
Paper analysisRead analysis
AttacksarXiv 2610.07645Oct 6
SkillPoison：用成功经验投毒自进化 Agent 的技能形成
提出 SkillPoison，用局部有效的成功经验诱导自进化 Agent 形成过度泛化技能
arXiv
2610.07645
Published
Oct 6
Layer
Application layer
Setting
AI Agent
Evaluated models
deepseek-v4-flash, Qwen V3.8 Flash, GPT-5.5
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
- ComponentMemory
AbstractSkillPoison 通过局部有效且验证成功的经验诱导技能过度泛化，在多个评测基准上实现了持久的下游攻击影响。
SkillPoison 提出了一种针对自进化智能体经验到技能形成过程的攻击范式，通过组织已验证的成功经验诱导技能过度泛化。
Paper analysisRead analysis
AttacksarXiv 2610.07323Oct 6
ATLAS：用自适应信任域与主动学习搜索潜在对抗样本集
提出 ATLAS 用水平集估计恢复黑盒分类器的对抗输入集合
arXiv
2610.07323
Published
Oct 6
Layer
Prompt layer
Setting
Models and APIs
Attacker
Black-box
Evaluated models
LeNet5, ResNet50 (CIFAR-10 standard), ResNet50 (ImageNet standard) and others5
- AttacksDetection evasion
- TechniqueAutomated search
AbstractATLAS 用主动水平集估计构造黑盒对抗集，作者称其集合比单点攻击更具代表性。
ATLAS 是一个查询式 black-box 框架，用来构造诱发失败的对抗输入集合，供鲁棒性审计使用。
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 2610.06049Oct 5
研究者提出仅修改 SAE 解码器的后门攻击，可在不改动语言模型的情况下植入行为
提出只改 SAE 解码器的后门，在冻结语言模型上植入代码插入
arXiv
2610.06049
Published
Oct 5
Layer
Model layer
Setting
Models and APIs
Evaluated models
Gemma 3 4B Instruct, Qwen3 1.7B, Llama 3.1 8B Instruct
- AttacksPoisoning and backdoors
- TechniqueDormant triggers
Abstract只改解码器的 SAE 可带上代码插入后门，且若干常规质量指标变化不大。
只改 SAE 解码器、冻结编码器与语言模型，即可把代码插入行为放进推理时的辅助组件。作者把这视为 SAE 供应链问题：checkpoint 一旦替换某一层激活，就参与后续计算，而不只是解释工具。
Paper analysisRead analysis
AttacksarXiv 2610.05894Oct 5
研究者提出闭环偏见注入攻击
提出闭环激活引导，在扩散语言模型去噪过程中注入定向偏见
arXiv
2610.05894
Published
Oct 5
Layer
Model layer
Setting
Models and APIs
Attacker
Gray-box
Evaluated models
LLaDA-8B-Instruct, Dream-v0-Instruct-7B, LLaDA-MoE-7B-A1B-Instruct
- AttacksModel tampering
- TechniqueGradient and representation optimization
Abstract通过去噪中间概率反馈动态调节残差流转向，证明扩散语言模型在推理服务栈中存在闭环偏见注入脆弱性。
Paper analysisRead analysis
DefensesarXiv 2610.05870Oct 5
MBZUAI 提出可校准的真实图像认证方法
提出校准重合成方法，以生成器反演保真度认证真实图像并防御对抗扰动
arXiv
2610.05870
Published
Oct 5
Layer
Application layer
Setting
LLM applications
Attacker
White-box
- DefensesDetection and filtering
- AttacksDetection evasion
- ComponentImage generation
Abstract论文提出基于反演保真度与阈值校准的健全认证框架，在对抗扰动下维持错误率界限，并指出真实内容可认证性随生成器演进而萎缩。
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 2610.06977Oct 4
IAU-FOA：面向闭源多模态大模型的可迁移对抗攻击方法
提出 IAU-FOA，构造可迁移图像扰动以诱导多模态模型输出目标语义
arXiv
2610.06977
Published
Oct 4
Layer
Prompt layer
Setting
Models and APIs
Evaluated models
Qwen2.5-VL-7B, Qwen2.5-VL-72B, InternVL3-14B and others17
- TechniqueGradient and representation optimization
- ComponentVision
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
DefensesarXiv 2610.03498Oct 2
检测与抑制：针对 VLA 模型对抗补丁的机制可解释性防御
提出条件特征抑制防御，检出对抗补丁后抑制 VLA 内部特征
arXiv
2610.03498
Published
Oct 2
Layer
Model layer
Setting
Embodied systems and robotics
Evaluated models
π0.5, SmolVLA
- DefensesInference-time intervention
- ComponentVision
Abstract条件抑制与 patch 相关的 SAE 特征，间歇攻击下可提高 SR，持续攻击下则不能。
Paper analysisRead analysis
AttacksarXiv 2610.03166Oct 2
LiBRA：通过双向隐空间优化实现检测感知的图像水印去除
提出 LiBRA，在潜空间双向优化以规避图像水印双侧检测
arXiv
2610.03166
Published
Oct 2
Layer
Model layer
Setting
Models and APIs
Attacker
White-box
Evaluated models
Stable Signature, YU1, HiDDeN and others4
- AttacksDetection evasion
- TechniqueDefense-targeted
AbstractLiBRA 用双向潜空间优化规避双侧水印检测，并提高相对反转基线的保真。
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

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

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.

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

研究者提出利用祖先 VLM 的黑盒对抗补丁攻击，可降低 VLA 任务成功率

P-VMI：对抗图像经 KV cache 持续影响多轮对话

STCA：研究者提出针对自动驾驶视频 VLM 的时空一致性对抗攻击

CCDF dataset: Grok Imagine, Veo 3.1, Sora 2 forge surveillance footage, ten deepfake detectors basically cannot recognize it

SkillPoison：用成功经验投毒自进化 Agent 的技能形成

ATLAS：用自适应信任域与主动学习搜索潜在对抗样本集

BabelFake：面向多语言音视频 DeepFake 检测的基准数据集

研究者提出仅修改 SAE 解码器的后门攻击，可在不改动语言模型的情况下植入行为

研究者提出闭环偏见注入攻击

MBZUAI 提出可校准的真实图像认证方法

研究者训练出按多智能体拓扑触发的代码后门模型

IAU-FOA：面向闭源多模态大模型的可迁移对抗攻击方法

Thrive：面向自回归 TTS 的抗重建攻击多比特语音水印框架

检测与抑制：针对 VLA 模型对抗补丁的机制可解释性防御

LiBRA：通过双向隐空间优化实现检测感知的图像水印去除

研究揭示无触发投毒审计缺口：事实答对不等于决策正确