建立AI準則與管制架構：歐盟《人工智慧法》的發展

Establishing AI guidelines and regulatory framework: The development of the EU Artificial Intelligence Act

近年來，人工智慧（AI）發展迅速，應用範圍日益擴大，成為全球政策與法規討論的重點。歐洲議會與歐盟理事會於2023年12月就全球首部《人工智慧法》（Artificial Intelligence Act）草案達成共識，並於2024年5月21日正式通過，旨在確保歐盟內可信賴的AI系統發展與單一市場運作。
該法案採風險導向管理（risk-based approach），依據風險程度將AI系統分類為「不可接受風險」、「高度風險」、「有限風險」與「低度風險」，並對應不同管制方式。隨著生成式AI技術進步，雖帶來產業與經濟新機會，但同時也引發資訊誤傳、人類安全等挑戰，對現行法規構成衝擊。
本文將深入分析《人工智慧法》的內容，探討其立法目的、AI管制架構及可能爭議，進一步探討建立AI準則的依據，並研析台灣未來立法的借鏡與建議。

In recent years, artificial intelligence (AI) has developed rapidly, with its applications expanding across various fields, making it a focal point of global policy and regulatory discussions. In December 2023, the European Parliament and the Council of the European Union reached a consensus on the draft of the world’s first Artificial Intelligence Act, which was officially passed on May 21, 2024. The act aims to ensure the development of trustworthy AI systems within the EU while maintaining the smooth functioning of the single market.
The legislation adopts a risk-based approach, categorizing AI systems into four levels according to their risk assessment: ''Unacceptable Risk,'' ''High Risk,'' ''Limited Risk,'' and ''Minimal Risk,'' each subject to corresponding regulatory measures. While the rapid advancement of generative AI has created new opportunities for industries and economic growth, it has also raised concerns regarding misinformation, human safety, and challenges to traditional regulatory frameworks.
This paper analyzes the Artificial Intelligence Act, examining its legislative intent, AI regulatory framework, and potential controversies. Furthermore, it explores the rationale behind establishing AI governance standards and provides insights into Taiwan’s future legislative considerations.