AI 輔助決策中的身心健康影響
Operation & Monitoring
Risk Description
Inappropriate AI responses in health, psychological counseling, or lifestyle advice scenarios may cause physical and psychological harm to users. For example, mental health chatbots giving inappropriate advice may exacerbate users psychological distress, fitness AI recommending overly intense exercise plans may cause physical injury, and AI companion apps may affect users normal social abilities and psychological health.
Framework Mappings
iso 23894R6
tw principle永續發展與福祉
tw principle人類自主
tw risk type部署互動問題（Deployment Interaction）
Controls English translation pending
- 健康建議安全護欄
- 專業轉介機制
Implementation Steps English translation pending
- 對涉及身心健康的 AI 回應設立嚴格的安全護欄
- 建立自動化的危機偵測和專業轉介機制
- 限制 AI 在心理健康場景中的角色（提供資訊而非治療）
- 定期評估 AI 健康建議的安全性和準確性