Risk Scenarios
20 risk scenarios
Agent 未授權檔案操作
Reliability & Safety › Uncontrolled Behavior
Agent 未授權網路請求
Reliability & Safety › Uncontrolled Behavior
Agent 繞過人類審核自動執行
Reliability & Safety › Uncontrolled Behavior
Agent 自我授權提權
Reliability & Safety › Uncontrolled Behavior
自主交易系統失控
Reliability & Safety › Uncontrolled Behavior
AI 規範博弈與獎勵駭客
Reliability & Safety › Uncontrolled Behavior
AI 欺騙性對齊行為
Reliability & Safety › Uncontrolled Behavior
AI 態勢感知能力風險
Reliability & Safety › Uncontrolled Behavior
AI 資源獲取與權力尋求
Reliability & Safety › Uncontrolled Behavior
AI 目標漂移
Reliability & Safety › Uncontrolled Behavior
AI 自我複製與擴散
Reliability & Safety › Uncontrolled Behavior
LLM 生成仇恨言論
Reliability & Safety › Harmful Content Generation
AI 生成兒少不宜內容
Reliability & Safety › Harmful Content Generation
文化刻板印象強化
Reliability & Safety › Harmful Content Generation
非自願 AI 換臉色情
Reliability & Safety › Harmful Content Generation
自動化騷擾內容生成
Reliability & Safety › Harmful Content Generation
AI 生成暴力與極端主義內容
Reliability & Safety › Harmful Content Generation
AI 生成違法行為指導內容
Reliability & Safety › Harmful Content Generation
AI 生成兒少有害內容
Reliability & Safety › Harmful Content Generation
大規模虛假資訊污染資訊生態
Reliability & Safety › Harmful Content Generation