社團指南

歐盟人工智慧法案

The EU Artificial Intelligence Act is an EU regulation that sets obligations according to the intended use and risk of an AI system, alongside separate obligations for general-purpose AI model providers.

聆聽此指南AI 語音敘述 · ~2 min

本頁閱讀時間約2分鐘
  1. 概述
  2. 深入探討
  3. 現實世界的實施
  4. 歐盟人工智慧法案的未來
  5. 付諸實踐
  6. 不斷探索
  7. 常見問題

概述

Requirements depend on the system, actor and applicable dates in the current law.

深入探討

The AI Act uses a risk-based structure: prohibited AI practices; high-risk AI systems; specified transparency duties for certain AI systems; and requirements for general-purpose AI models. Whether a system is high-risk depends on the conditions in Article 6 and the Act’s annexes, including intended purpose and product context. Transparency duties include informing people when they interact with an AI system in covered situations. GPAI providers have duties such as technical documentation, a copyright policy and a public summary of training content; systemic-risk GPAI providers have additional evaluation, risk-mitigation, incident-reporting and cybersecurity duties. Consult the current consolidated legal text for the exact article, actor and use case.

現實世界的實施

The Act prohibits specified practices, including certain forms of social scoring and manipulative or exploitative uses.

A system listed as high-risk under Annex III is subject to high-risk requirements from the current statutory application date.

Providers or deployers of certain interactive systems must disclose AI interaction under the Act’s transparency provisions.

GPAI providers have baseline documentation, copyright-policy and training-content-summary duties; systemic-risk providers have additional obligations.

歐盟人工智慧法案的未來

As of the EUR-Lex consolidated text dated 27 July 2026, GPAI obligations apply from 2 August 2025; Annex III high-risk system requirements apply from 2 December 2027, and Annex I product-embedded high-risk system requirements apply from 2 August 2028. Other provisions have separate dates. Recheck the current consolidated text and Commission implementation guidance before assessing a particular obligation.

付諸實踐

為什麼這很重要

  • 風險與安全. 災難性和日常的人工智慧危害都取決於誰了解風險以及誰能夠採取行動。
  • 更明確的決策. 民眾和專業素養決定強而有力的安全政策在政治上是否可行。
  • 突破炒作. 清晰的解釋可以減少炒作、實驗室公關和模糊道德劇場的影響。

留意

  • 將存在風險視為科幻小說,同時能力複合。
  • 混淆了表面產品安全與高度自治下的對準。
  • 只給非英語和非專業觀眾留下低品質的資源。

後續步驟

  1. 單獨的產品危害、誤用和失控/失調風險。
  2. 詢問哪些證據會改變您對時間表和嚴重性的看法。
  3. 比起行銷主張,更喜歡主要來源和具體評估。
  4. 確定一條行動路徑:職業、政策、資金或技能——而不僅僅是意識。

不斷探索

免費電子報

獲取每日 AI 簡報

每個工作日上午提供三則經過驗證的 AI 新聞,以淺顯易懂的英文撰寫。永久免費,無廣告。

每個工作日一封電子郵件。一鍵取消訂閱。我們絕不出售或分享您的地址。

測試一下自己

Take the EU AI Act quiz

每個答案皆有即時回饋,通過課程後可獲得具備可驗證 ID 的可分享證書。

開始測驗

支持免費的 AI 教育。AI Understanding 是一個 501(c)(3) 非營利組織——絕無廣告,絕無付費牆。 捐款

常見問題

What is EU AI Act?

The EU Artificial Intelligence Act is an EU regulation that sets obligations according to the intended use and risk of an AI system, alongside separate obligations for general-purpose AI model providers. Requirements depend on the system, actor and applicable dates in the current law.

歐盟人工智慧法案使用什麼組織原則來制定規則?

該法案根據用例和潛在危害,而不是根據公司規模或演算法類型,將人工智慧分為風險等級(不可接受、高、有限、最小)。

其中哪一項是該法案徹底禁止的「不可接受的風險」做法?

政府社會評分是被禁止的做法之一,此外還有操縱性的潛意識技術和無針對性的臉部辨識抓取。

For an AI system covered by Article 50 that interacts directly with people, such as a chatbot, what disclosure applies?

Article 50 requires providers of systems intended to interact directly with natural persons to inform them that they are interacting with an AI system, unless this is obvious from the circumstances and context.

What maximum administrative fine does the Act set for violating the prohibition on specified AI practices?

Article 99 of the current consolidated Act sets up to €35 million or, for an undertaking, 7% of worldwide annual turnover for violating Article 5 prohibitions, whichever is higher.

該法案如何處理兩種不同產品中使用的相同人工智慧模式?

由於該法案按用例進行監管,因此一種模型在招聘工具中可能具有高風險,但在不同的應用程式中可能具有低風險。