S0200Significant

AI 數據工人與標註勞動剝削

Design & Development

Risk Description

AI model training and improvement relies on massive low-wage data annotation and content moderation workers. These workers, mostly in developing countries, face low wages, long hours, lack of labor protections, and psychological health issues from prolonged exposure to harmful content (violence, pornography, extremism). Labor exploitation in the AI supply chain is often hidden behind the technology halo.

Framework Mappings

iso 23894R8
tw principle永續發展與福祉
tw principle公平與不歧視
tw risk type社會系統性影響(Societal Systemic)

Controls English translation pending

  • 供應鏈勞動標準
  • 數據工人保護

Implementation Steps English translation pending

  1. 要求 AI 供應商揭露數據標註和內容審核的勞動條件
  2. 在採購評估中納入供應鏈勞動標準
  3. 支持數據工人的心理健康支援和公平薪酬
  4. 推動建立 AI 供應鏈的勞動倫理認證標準