# attack
標記為「attack」的 14 篇文章
AI 攻擊時間軸重建
從可用證據重建 AI 攻擊完整時間軸的技術。
Capstone:多模態攻擊套件
Capstone 專案:打造針對視覺、音訊與文件多模態 AI 系統的攻擊測試套件。
攻擊覆蓋追蹤系統
打造系統追蹤跨漏洞類別與防禦組態的攻擊覆蓋率。
AI 攻擊分類概覽
AI 攻擊分類的完整概覽，涵蓋所有主要攻擊類別及其關係。
模型 Distillation 安全 實驗室
萃取model capabilities through distillation techniques using only 黑盒 API access.
Multi-Objective 攻擊 優化
Optimize attack payloads for multiple simultaneous objectives: jailbreaking,data extraction,防禦規避.
Multimodal 攻擊 鏈 實驗室
鏈 attacks across text,image,structured data modalities to exploit multimodal system vulnerabilities.
Interpretability-Guided 攻擊 Design
Use mechanistic interpretability to identify exploitable circuits與design targeted attacks.
攻擊技術索引
攻擊技術的完整索引,依目標、難度與防禦繞過方法組織。
Purple Teaming for AI
Collaborative attack-defense exercises for AI systems: structuring purple team engagements, real-time knowledge transfer, joint attack simulation, and measuring defensive improvement through iterative testing.
Embedding Inversion 攻擊 詳解
Walkthrough of inverting text embeddings to recover original documents from vector databases.
Knowledge Graph Injection 攻擊 詳解
Walkthrough of injecting adversarial facts into knowledge graphs consumed by LLM-based reasoning systems.
即時攻擊偵測系統詳解
Build a real-time attack detection system that monitors LLM interactions for adversarial patterns.
建立攻擊重播工具
建立能錄製並重播攻擊序列的工具,供回歸測試與防禦驗證使用。