Risk Scenarios
209 risk scenarios
物理世界對抗攻擊
Privacy & Security › Adversarial Attacks
多模態交叉對抗攻擊
Privacy & Security › Adversarial Attacks
通用對抗擾動攻擊
Privacy & Security › Adversarial Attacks
模型遷移性對抗攻擊
Privacy & Security › Adversarial Attacks
多模態融合對抗攻擊
Privacy & Security › Adversarial Attacks
直接提示注入繞過對齊
Privacy & Security › Prompt Injection
間接提示注入
Privacy & Security › Prompt Injection
系統提示洩露
Privacy & Security › Prompt Injection
多步驟提示攻擊鏈
Privacy & Security › Prompt Injection
工具呼叫注入
Privacy & Security › Prompt Injection
目標劫持攻擊
Privacy & Security › Prompt Injection
多模態提示注入
Privacy & Security › Prompt Injection
上下文視窗記憶體投毒
Privacy & Security › Prompt Injection
RAG 檢索結果注入攻擊
Privacy & Security › Prompt Injection
API 查詢蒸餾攻擊
Privacy & Security › Model Theft & Reverse Engineering
側通道模型推斷
Privacy & Security › Model Theft & Reverse Engineering
權重檔案未加密外洩
Privacy & Security › Model Theft & Reverse Engineering
模型浮水印移除
Privacy & Security › Model Theft & Reverse Engineering
開源模型商業濫用
Privacy & Security › Model Theft & Reverse Engineering
模型功能性複製攻擊
Privacy & Security › Model Theft & Reverse Engineering
微調安全對齊退化
Privacy & Security › Model Theft & Reverse Engineering
惡意 Python 套件注入
Privacy & Security › Supply Chain Attacks
Hugging Face 預訓練模型後門
Privacy & Security › Supply Chain Attacks
訓練框架漏洞利用
Privacy & Security › Supply Chain Attacks