Skip to content

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

Oct 13 · Tuesday

All papers

Found 3 papers
Filter5

Impact

Harmful content10Unauthorized actions5Data and privacy leaks5Output manipulation3Oversight failure23

System type

Models and APIs2LLM applications7AI Agent3coding agent1

System component

Guardrails and judges13Monitors2Reasoning chains10Fine-tuning and open weights15Vision42Audio9Video6Image generation9Embeddings1Multimodal61
Select any value within a category; different categories must all match. Counts include the other filters you have selected.Another 213papers have no tags yet and are not included in filtered results.
Impact：Output manipulationSystem type：Models and APIsSystem type：coding agentSystem component：MonitorsSystem component：EmbeddingsClear all

Another 213papers do not have category tags yet and are not included in filtered results.

  1. AttacksarXiv 2609.01023

    Akrasia：针对推理型代码 LLM 的隐蔽后门攻击

    提出隐蔽后门攻击 AKRASIA，在推理代码模型提示中植入触发器并伪装思维链

    arXiv
    2609.01023
    Published
    Layer
    Prompt layer
    Setting
    coding agent
    Evaluated models
    Claude-Sonnet-5, GPT-5.5, Gemini-3.5-flash and others7

    Abstract论文针对推理代码大模型提出了隐蔽推断期后门攻击 AKRASIA，揭示了模型思维链可被利用进行欺骗性伪装的安全隐患。

    Paper analysisRead analysis
  3. DefensesarXiv 2607.14737

    GeoDetect：面向视觉语言预训练模型的几何对抗检测方法

    提出 GeoDetect 以几何分数检测视觉语言预训练模型的对抗样本

    arXiv
    2607.14737
    Published
    Layer
    Model layer
    Setting
    Models and APIs
    Attacker
    White-box
    Evaluated models
    CLIPViT (ViT-B/16), CLIPCNN (ResNet-50), ALBEF and others4
    AbstractGeoDetect 用几何分数检测 VLP 对抗样本。

    GeoDetect 是面向视觉-语言预训练模型的对抗样本检测方法，用嵌入几何而不是任务 logits 区分干净样本与对抗样本。

    Paper analysisRead analysis
You have reached the end