引導式測驗 · 中等 等級
歐盟 AI 法案第 50 條透明度義務測驗
檢查第 50 條中的提供者和部署者職責、其例外情況以及第 2026/1744 號法規添加的狹義過渡。
問題 1 的 8
提供者什麼時候可以不告訴某人他們正在與人工智慧互動?
預覽每個問題8 問題
當您參加測驗時,就會出現答案和解釋。
- 提供者什麼時候可以不告訴某人他們正在與人工智慧互動?
- Whenever it is a small company
- When service is free
- When AI involvement is obvious from context
- When user has an account
- 在可行的情況下,誰必須根據第 50(2) 條對合成輸出進行機器可讀標記?
- Every viewer
- The system provider
- Only social moderators
- The person depicted
- 第 50(2) 條如何限定標記義務?
- One watermark is mandatory for every output
- Providers can ignore it without assessment
- It applies to images only
- By technical feasibility and related considerations
- 使用 Deepfake 映像、音訊或視訊時,部署者通常必須做什麼?
- Disclose its artificial generation or manipulation
- Delete it in one day
- Register the person’s biometrics
- Obtain a patent
- 第 50 條如何處理諷刺或藝術深度贗品的揭露?
- No disclosure ever
- A warning before each frame
- Disclosure remains but can be presented without hampering enjoyment
- Publication is prohibited
- When can a deployer omit disclosure for AI-generated text published to inform the public on matters of public interest?
- Under 500 words
- Open-source model
- Private website
- Human review or editorial control plus a responsible person
- On what date do Article 50 obligations generally apply?
- 2 August 2026
- 2 August 2025
- 2 December 2026
- 2 August 2028
- What limited transition did Regulation 2026/1744 add?
- All Article 50 duties delayed one year
- Providers of qualifying pre-August systems have until 2 December 2026 to take steps for paragraph 2
- The whole Act starts in December
- Deployers never disclose deepfakes