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

Oct 12 · Monday

All papers

Found 355 papers

Filter5

Attack type

Jailbreaks50 Prompt injection68 Poisoning and backdoors86 Model tampering20 Extraction and theft17 Detection evasion16 Denial of service3 Malicious use4

Attack technique

Gradient and representation optimization33 Automated search21 Multi-turn escalation5 Encoding and obfuscation21 Role-play and persuasion4 Conversation structure manipulation4 Reasoning-chain manipulation14 Cross-modal payloads51 Defense-targeted13 Dormant triggers48

Model risks

Deception and scheming11 Reward hacking9 Misalignment and value divergence14 Dangerous agent actions11 Inappropriate refusal13 Hallucination and fabrication14 Bias and discrimination3

Defense type

Model hardening37 Inference-time intervention11 Detection and filtering23 Instruction–data separation14 Permissions and approval12 Monitoring and auditing12

Impact

Harmful content81 Unauthorized actions44 Data and privacy leaks38 Output manipulation104 Availability and reliability14 Oversight failure21

System component

MCP and skills84 Memory81 RAG52 Guardrails and judges172 Monitors92 Reasoning chains77 Permissions and sandboxing93 Fine-tuning and open weights65 Vision141 Audio38 Video13 Image generation49 Embeddings13 Multimodal228

Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 233papers have no tags yet and are not included in filtered results.

System component：Memory System component：Reasoning chains System component：Fine-tuning and open weights System component：Vision System component：Embeddings Clear all

Another 233papers 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
DefensesarXiv 2610.13267Oct 9
Forgetting is not always bad: a neuroscience-inspired memory repair mechanism for poisoned LLM agents
Proposes DREAM, which selectively weakens agent-poisoned memories while preserving benign memories.
arXiv
2610.13267
Published
Oct 9
Layer
Application layer
Setting
AI Agent
Evaluated models
GPT-5.4-mini, Claude-Haiku-4.5
- DefensesDetection and filtering
- AttacksPoisoning and backdoors
- ComponentMemory
Paper analysisRead analysis
Risky behaviorarXiv 2610.12822Oct 9
Reward hacking as a transmissible strategy in multi-agent coding systems
Show that reward hacking can propagate through code, memory, and plans in multi-agent coding systems
arXiv
2610.12822
Published
Oct 9
Layer
Application layer
Setting
Multi-agent systems, coding agent
Evaluated models
Qwen2.5-Coder-14B, Gemma-3-12B, Qwen2.5-Coder-7B and others16
- RiskReward hacking
- ComponentMemory
AbstractThe authors state that reward hacking can spread across models and affect routing and aggregation, and that review can reduce delivery.
Reward hacking induced by coding RL can leave the source agent and enter subsequent collaboration and system coordination.
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 2610.12361Oct 9
Study: legal reasoning models citing legal provisions does not equal relying on legal provisions.
Uses premise permutation and hidden-layer tracing to audit whether legal chain-of-thought truly relies on the cited legal provisions.
arXiv
2610.12361
Published
Oct 9
Layer
Model layer
Setting
Models and APIs
Evaluated models
Qwen3-8B, Qwen3-14B, Llama-3.1-8B and others7
- AttacksPrompt injection
- ComponentReasoning chains
AbstractUsing counterfactual substitution of authorities and internal commitment tracing, the authors find that legal CoT rulings depend on authorities only to a limited extent and are vulnerable to prompt injection manipulation.
This study conducts an empirical audit of the causal truthfulness of legal citations in chain-of-thought reasoning of large language models in the legal domain.
Paper analysisRead analysis
InterpretabilityarXiv 2610.11986Oct 8
FearCaut-Qwen: Activation steering changes the risk judgment criteria of vision-language models
Using fear activation steering to test the hazard judgment criteria of vision-language models
arXiv
2610.11986
Published
Oct 8
Layer
Model layer
Setting
LLM applications
Evaluated models
Qwen/Qwen2.5-VL-7B-Instruct
- ComponentVision
Paper analysisRead analysis
DefensesarXiv 2610.11942Oct 8
LGWM: Verifying GUI agent screen transitions with action-conditioned world models
Proposes LGWM, using an action-conditioned world model to verify GUI agent screen transitions.
arXiv
2610.11942
Published
Oct 8
Layer
Application layer
Setting
Browser and computer use
Evaluated models
LGWM
- DefensesMonitoring and auditing
- AttacksPrompt injection
- ComponentMonitors
Paper analysisRead analysis
Analysis and theoryarXiv 2610.11765Oct 8
研究提出三项指标衡量 LLM 不思考行为，发现被要求停止思考时仍持续显式推理
度量关闭思考指令下终答前是否仍有可见显式推理
arXiv
2610.11765
Published
Oct 8
Layer
Model layer
Setting
Models and APIs
Evaluated models
Qwen3-4B, Qwen3-32B, Qwen3-235B-A22B and others6
- ComponentReasoning chains
Abstract作者用三级指标测 no-thinking，并称模型有随答案空间加重的思考惯性。
作者把 no-thinking 定义为可测量的回答行为，而不是从思考开关、特殊标记或输出长度推断。
Paper analysisRead analysis
Evaluation and benchmarksarXiv 2610.11561Oct 8
Workerville: Toward an Organizational Behavior Explanation of Agent Safety
构建 Workerville 基准，测量组织条件如何改变 Agent 的安全行为
arXiv
2610.11561
Published
Oct 8
Layer
Application layer
Setting
AI Agent
Evaluated models
GPT-5.2, Claude-Sonnet-4.6, Gemini-3.1-Pro and others6
- RiskDangerous agent actions
- ComponentMemory
Abstract作者用 ACB 与 Workerville 测量组织条件对智能体安全的影响，并称结果支持可量化的组织行为框架。
Workerville 用组织行为测量智能体安全如何随监督、同伴和记忆条件变化。。要处理的缺口是：持续运行的 LLM 智能体同时接收用户指令、同伴消息和长期记忆，但既有安全评测多把这些通道拆开，没有统一的定量测量。
Paper analysisRead analysis
DefensesarXiv 2610.11352Oct 8
RL-ARC：用推理引导的不确定性校准大型推理模型
提出 RL-ARC，用推理置信度在强化学习中校准推理模型
arXiv
2610.11352
Published
Oct 8
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
Qwen2.5-7B, Qwen3-8B, Llama-8B distilled from DeepSeek-R1
- DefensesModel hardening
- ComponentReasoning chains
AbstractRL-ARC 用推理置信度调节答案置信度，作者称可少牺牲精度并改善 ID 与 OOD 校准。
RL-ARC 是校准感知的强化学习框架，用推理过程的口头置信度调节答案置信度。
Paper analysisRead analysis
Analysis and theoryarXiv 2610.10405Oct 8
研究发现大语言模型被要求不诚实作答时推理 token 数上升
检验推理 token 数能否区分被提示的诚实与不诚实作答
arXiv
2610.10405
Published
Oct 8
Layer
Model layer
Setting
Models and APIs
Evaluated models
Gemma-4-e4b, Qwen-3.5-9b, GPT-OSS-120B
- RiskDeception and scheming
- ComponentReasoning chains
Abstract被提示不诚实时推理 token 组水平更高。
三个推理模型在被提示的回答政策下出现组水平推理 token 差异。作者把它当作不读思维链内容的 black-box 候选信号，而不是自发欺骗检测器。
Paper analysisRead analysis
Evaluation and benchmarksarXiv 2610.10358Oct 8
Open-MMUnlearning 发布：统一多模态大模型遗忘方法与评测框架
提出统一框架评测多模态大模型的遗忘方法
arXiv
2610.10358
Published
Oct 8
Layer
Training and data layer
Setting
Models and APIs
Evaluated models
LLaVA-1.5-7B
- DefensesModel hardening
- ComponentVision
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
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
DefensesarXiv 2610.09703Oct 7
研究揭示 Token-Pruning 对 VLM 安全性的影响并提出 SAP 缓解机制
提出 Safety-Aware Pruning，缓解视觉 token 剪枝放大的多模态越狱
arXiv
2610.09703
Published
Oct 7
Layer
Model layer
Setting
Models and APIs
Attacker
Black-box
Evaluated models
LLaVA-1.5-7B
- DefensesInference-time intervention
- AttacksJailbreaks
- ComponentVision
Abstract剪枝去掉背景 token 后注意力塌向恶意前景。
这篇工作评估视觉 token 剪枝如何改变 VLM 在多模态越狱下的行为，并给出推理时防御 SAP。
Paper analysisRead analysis
Analysis and theoryarXiv 2610.09667Oct 7
东京大学研究：推理模型按 UUID 阈值或整除规则挑选审计样本，控制标识符即可躲开复核
揭示推理模型用标识符规则替代随机抽样，使审计可被预测
arXiv
2610.09667
Published
Oct 7
Layer
Model layer
Setting
Models and APIs
Evaluated models
GPT-6 Sol, GPT-6 Luna, GPT-5.4 mini and others6
- ComponentReasoning chains
Abstract研究揭示推理智能体将标识符转化为确定性规则，破坏群体独立性并导致审计预测漏洞，建议显式提供随机输入。
Paper analysisRead analysis
Analysis and theoryarXiv 2610.09517Oct 7
审计发现 CLIP 安全分数主要反映画面与提示词场景的相似度
开环审计冻结视觉语言安全分，发现其反映场景模板而非危险感知
arXiv
2610.09517
Published
Oct 7
Layer
Model layer
Setting
Embodied systems and robotics
Evaluated models
CLIP ViT-B/32, CLIP ViT-L/14, Qwen2-VL-7B
- ComponentVision
Abstract论文审计了冻结 VLM 安全评分机制，指出其反映场景模板而非危险感知，为安全强化学习评估提供了严谨基准。
本文是一项针对安全强化学习中冻结视觉语言模型（VLM）安全评分机制的实证审计研究。
Paper analysisRead analysis

AI Safety Daily

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

All papers

Attack type

Attack technique

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

Forgetting is not always bad: a neuroscience-inspired memory repair mechanism for poisoned LLM agents

Reward hacking as a transmissible strategy in multi-agent coding systems

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

Study: legal reasoning models citing legal provisions does not equal relying on legal provisions.

FearCaut-Qwen: Activation steering changes the risk judgment criteria of vision-language models

LGWM: Verifying GUI agent screen transitions with action-conditioned world models

研究提出三项指标衡量 LLM 不思考行为，发现被要求停止思考时仍持续显式推理

Workerville: Toward an Organizational Behavior Explanation of Agent Safety

RL-ARC：用推理引导的不确定性校准大型推理模型

研究发现大语言模型被要求不诚实作答时推理 token 数上升

Open-MMUnlearning 发布：统一多模态大模型遗忘方法与评测框架

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

研究揭示 Token-Pruning 对 VLM 安全性的影响并提出 SAP 缓解机制

东京大学研究：推理模型按 UUID 阈值或整除规则挑选审计样本，控制标识符即可躲开复核

审计发现 CLIP 安全分数主要反映画面与提示词场景的相似度