一、前言1. Preface
有見於生成式人工智能技術迅速發展，並在教育領域具有廣泛應用潛力，本指引旨在為本校師生提供清晰的行為規範與建議，確保人工智能工具能被負責任、合乎道德且有效地運用，同時維護學術誠信與個人資料安全。
生成式人工智能服務須履行資訊透明的責任，讓使用者明白其能力與限制。本指引亦旨在協助香港教育科技領域確保資源公平可及，避免學習成果因演算法偏見而受影響。
In light of the rapid development of generative AI technology and its potential applications in education, these guidelines aim to provide clear behavioural norms and recommendations for school teachers and students. The goal is to ensure that AI tools are used responsibly, ethically, and effectively, while upholding academic integrity and protecting personal data.
Generative AI services bear the responsibility of transparency, enabling users to understand their capabilities and limitations. These guidelines are also intended to help ensure equitable access in the Hong Kong education technology sector, preventing learning outcomes from being affected by algorithmic bias.
二、學生版指引2. Guidelines for Students
學生在使用生成式人工智能時應保持警惕，並承擔「人工智能有益向善的主動維護者」的責任——即當發現 AI 所生成的內容具誤導性、虛假或有害時，應主動指出而非散播。
本校並不一律禁止學生使用生成式人工智能，但學生必須學習如何負責任地運用此工具。
為幫助學生具體判斷使用時機，本校引入「AI 學習紅綠燈應用框架」。學生在課業中使用 AI 前，應以此框架自行評估；若有任何疑問，必須主動向教師徵詢意見並取得同意。
When using generative AI, students should remain vigilant and assume the responsibility of being proactive stewards of AI's beneficial and ethical development—that is, when AI-generated content is found to be misleading, false, or harmful, students should actively point this out rather than spread it further.
The school does not universally prohibit students from using generative AI; however, students must learn to use this tool responsibly.
To assist students in making specific determinations regarding the appropriate scenarios for AI utilisation, the school has implemented the "Traffic Light Application Framework for AI in Learning". Prior to employing AI in their coursework, students are required to utilise this framework for self-assessment. Should any uncertainty arise, they must proactively seek guidance and obtain approval from teachers.
三、AI 學習「紅綠燈」應用框架3. The "Traffic Light" Framework
如果對某情境屬於紅燈、黃燈或綠燈感到不確定，請預設按紅燈處理——先暫停使用，並主動向科任教師查詢，待獲得明確指示後方可繼續。
If you are unsure whether a situation falls under Red, Yellow, or Green Light, default to Red Light—pause, then consult your subject teacher and proceed only upon receiving clear guidance.
點擊下方任一燈色，了解該類情境的定義與範例：
Click any traffic light below to see its definition and examples:
🚦 紅燈：禁止使用
🚦 Red Light: Prohibited Use
- 正式考試、測驗或課堂小測。
- Formal examinations, tests, or in-class quizzes.
- 要求展現個人經驗、觀點反思或創意寫作的作業，例如個人陳述、讀書報告的核心評論等。
- Assignments requiring personal experience, critical reflection, or creative writing—e.g., personal statements, core analysis in a book report.
- 教師明確規定須獨立完成的解題過程或學術分析。
- Problem-solving processes or academic analyses where the teacher has explicitly required independent work.
🚦 黃燈：謹慎使用並必須註明
🚦 Yellow Light: Use with Caution and Citation
- 腦力激盪：尋求主題靈感或大綱建議，學生須自行重組並深化內容。
- Brainstorming: Generating ideas for a topic or outline, which the student must then restructure and develop.
- 資料搜集：協助整理初步資料或提供不同觀點，學生必須核實資料的準確性。
- Data Gathering: Assisting with preliminary information, which the student must fact-check.
- 文法潤飾：在完成初稿後，用以檢查文法、拼寫或改善句式。
- Proofreading: After your initial draft, AI may check grammar, spelling, or refine sentence structure.
- 程式除錯：協助找出程式碼中的錯誤，學生須理解並親自修正。
- Code Debugging: Identifying errors—students must comprehend and personally rectify them.
🚦 綠燈：鼓勵使用
🚦 Green Light: Encouraged Use
- 概念學習：請 AI 以簡單的比喻解釋複雜的科學或學術概念。
- Concept Learning: Ask AI to explain complex concepts using simple analogies.
- 個人化練習：根據個人進度生成練習題或詞彙表。
- Personalised Practice: Generate practice questions or vocabulary lists.
- 外語練習：進行模擬對話，提升口語與寫作流利度。
- Language Practice: Simulate conversations to improve fluency.
- 擴展視野：探索全新領域的基礎知識。
- Exploring New Fields: Gain foundational knowledge in a new topic.
- 整理筆記：將上課筆記或長篇文章整理成摘要與重點。
- Note Organisation: Organise notes or long texts into summaries.
- 自我測驗：請 AI 出題以考核自己對教材的掌握程度。
- Self-quizzing: Ask AI to generate quiz questions to test your understanding.
- 制定溫習計劃：訂立溫習時間表及學習目標。
- Study Planning: Draft study schedules and learning goals.
- 反向講解：將概念講解予 AI，由 AI 指出你的理解偏差。
- Reverse Explanation: Explain a concept to AI; let it spot your gaps.
- 外語材料的輔助翻譯：須先自行嘗試理解後，再使用 AI 對照核實。
- Assisted Translation: After attempting to understand foreign material yourself.
四、其他重要指引4. Other Important Guidelines
以下 10 項細則涵蓋學術誠信、私隱、知識產權、心理健康及外部考試等範疇。可逐項展開閱讀。
The following 10 items cover academic integrity, privacy, IP, mental health, and external exams. Expand each to read.
(一) 內容識別與學術誠信 (1) Content Identification and Academic Integrity
承接「黃燈」原則，應確保 AI 生成的內容能夠被明確識別。學生應清楚標明作業中哪些部分屬 AI 輔助完成（例如於附錄或註腳中說明），以避免出現有違學術誠信的情況。
當作業屬於黃燈情境時，請於作業末尾附上「AI 使用聲明」，使教師能清楚掌握 AI 輔助的範圍。下方為互動範本，填寫後可一鍵複製：
Following the "Yellow Light" principle, AI-generated content must be clearly identifiable. Students should clearly indicate which parts of their assignments were assisted by AI (e.g., in an appendix or footnote) to prevent breaches of academic integrity.
For Yellow Light situations, attach an "AI Usage Declaration" at the end of your assignment so teachers may understand the scope of AI assistance. Fill in the interactive template below and copy with one click:
(二) 防範 AI 偽造引用 (2) Verifying AI-Generated Citations
生成式人工智能經常會「捏造」出並不存在的參考文獻、書名、人物或統計數據——此現象稱為 AI hallucination（人工智能幻覺）。
凡由 AI 提供的引用、數據、人名、書名或網址，學生必須親自核實原始來源；無法查證者，不得寫入作業。
Generative AI frequently fabricates non-existent citations, book titles, persons, or statistics—this phenomenon is called AI hallucination.
For any citation, datum, name, title, or URL provided by AI, students must personally verify the original source. Anything that cannot be verified must not be included in the assignment.
(三) 影像、影片、聲音生成的使用守則 (3) Generative Visual, Video, and Audio Content
紅黃綠燈原則同樣適用於圖像（如 Midjourney、DALL-E）、影片（如 Sora、Runway）及聲音（如 ElevenLabs、Suno）等多模態生成式人工智能。
The Traffic Light framework also applies to multimodal generative AI for images (e.g., Midjourney, DALL-E), videos (e.g., Sora, Runway), and audio (e.g., ElevenLabs, Suno).
未經當事人明確同意，不得利用 AI 生成、合成或修改任何真實人物之肖像、聲音或影像（包括 deepfake、換臉、聲音模仿等技術）。如有違反，無論創作意圖為何，均屬嚴重違紀。
此類違規行為亦可能觸犯香港法律（例如《個人資料（私隱）條例》、《淫褻及不雅物品管制條例》及《刑事罪行條例》的相關條文）。
Without the explicit consent of the person concerned, no one may use AI to generate, synthesise, or alter the likeness, voice, or image of any real person (including deepfakes, face-swaps, voice cloning, and similar techniques). Any breach, regardless of intent, will be treated as serious misconduct.
Such breaches may also contravene Hong Kong law (e.g., the Personal Data (Privacy) Ordinance, the Control of Obscene and Indecent Articles Ordinance, and relevant provisions of the Crimes Ordinance).
(四) 堅持自主判斷與核實 (4) Maintaining Autonomous Judgement and Verification
- 生成式人工智能是學生的工具與助手，並非學生的替代者。AI 所生成的內容可能包含誤導性、虛假或不準確的資訊。
- AI 生成的內容可作為工作與學習的參考，但不應在未經驗證與判斷的情況下被採用。作為負責任的使用者，學生應具備法律、道德與風險管理等多方面的意識與能力，能透過驗證與審查作出獨立的資訊判斷。
- Generative AI is a tool and assistant, not a substitute. AI-generated content may contain misleading, false, or inaccurate information.
- AI-generated content may serve as a reference, but should not be adopted without verification and judgement. Students should possess awareness across legal, ethical, and risk-management aspects, and be capable of independent informational judgement.
(五) AI 與個人情緒、心理健康 (5) AI and Emotional / Mental Well-being
近年越來越多學生將生成式人工智能用作傾訴對象，或用以處理情緒困擾。本校謹此提醒：
- AI 並非專業輔導員。其回應基於統計模式，並未真正理解使用者的情緒、家庭與文化背景，亦無法承擔諮詢責任。
- 若感到持續低落、焦慮、無助，或出現傷害自己的念頭，請即時向學校社工、輔導員、班主任或可信任的家人尋求支援。
- 切勿將 AI 作為唯一傾訴對象或長期情感依賴。
An increasing number of students use generative AI as a conversational companion or for processing emotional concerns. The school reminds you that:
- AI is not a professional counsellor. Its responses are based on statistical patterns; it does not truly understand your emotions, family, or cultural context.
- If you experience persistent sadness, anxiety, helplessness, or thoughts of self-harm, please seek immediate support from a school social worker, counsellor, class teacher, or trusted family member.
- Do not rely on AI as your sole confidant or long-term emotional anchor.
(六) 小組功課與資料同意 (6) Group Work and Data Consent
- 將他人的作業、私訊或個人資料輸入 AI 工具前，須先取得當事人的同意。
- 在小組作業中，若任何組員曾使用 AI 輔助，須主動向所有組員披露，並反映於小組共同提交的 AI 使用聲明之中。
- Before inputting another person's work, private messages, or personal data into an AI tool, obtain that person's consent first.
- In group projects, any member who has used AI assistance must proactively disclose this to all teammates and reflect it in the group's collective AI Usage Declaration.
(七) 了解責任與義務 (7) Understanding Responsibilities and Obligations
- 在使用任何生成式人工智能服務之前，應仔細閱讀並熟悉相關平台或軟件的使用條款。
- 不得將相關服務用於違法、違規或不當的目的。例如，不得指使 AI 生成任何具有仇恨、歧視、誹謗或其他不道德及違法性質的內容。
- Before using any generative AI service, carefully read and familiarise yourself with the platform's terms of use.
- The service must not be used for illegal, non-compliant, or inappropriate purposes—e.g., generating hateful, discriminatory, defamatory, immoral, or illegal content.
(八) 保障個人私隱 (8) Protecting Personal Privacy
- 應熟悉所使用之生成式人工智能服務的私隱政策，以了解其資料收集、使用與共享的具體做法。
- 建議選用不會將共享資料用於 AI 訓練的服務，並避免傳輸敏感的個人資訊（例如身份證號碼、住址、個人病歷），同時可採用化名或匿名手段以保護私隱。
- Familiarise yourself with the privacy policies of the AI services you use.
- Choose services that do not use shared data for AI training; avoid transmitting sensitive personal info (ID numbers, home addresses, medical records). Pseudonymisation or anonymisation may also help.
(九) 謹慎傳播相關內容及尊重知識產權 (9) Cautious Dissemination and Respect for Intellectual Property
- 任何由生成式人工智能系統所生成並進一步散播的內容，最終責任仍由內容傳播者承擔。學生須主動檢查所生成內容的真實性、合法性與適宜性。
- 如擬公開發布 AI 生成的內容，應於公開時披露其來源。
- 自覺尊重知識產權，避免生成的內容構成對受版權保護作品的全部或實質性複製。
- The ultimate responsibility for AI-generated content that is further disseminated rests with the disseminator. Verify truthfulness, legality, and appropriateness.
- If you publish AI-generated content, disclose its source.
- Respect intellectual property; avoid generating content that constitutes a complete or substantial reproduction of copyrighted works.
(十) 升學、公開考試及外部評核 (10) Admissions, Public Exams, and External Assessments
學生於以下場合，須特別留意每一機構的具體 AI 政策；部分機構的要求遠較本校指引嚴格：
- DSE 校本評核（SBA）、獨立專題探究（IES）及其他公開考試的相關項目。
- 大學個人陳述、UCAS、Common App 及各地大學申請文件。
- 校外比賽、獎學金申請及學術論文投稿。
若目標機構並未明示其 AI 政策，預設按最嚴格標準（紅燈）處理，並向相關教師徵詢意見。
For these contexts, pay particular attention to each institution's specific AI policy—some impose stricter requirements:
- HKDSE SBA, IES, and other public-exam coursework.
- University personal statements, UCAS, Common App, and overseas applications.
- External competitions, scholarships, and academic paper submissions.
If the target institution has not published a clear AI policy, default to the strictest standard (Red Light) and consult the relevant teacher.
五、學生自我檢查表5. Student Self-check Checklist
每次使用 AI 之前，請自問以下五項問題。逐項剔選確認自己已準備好：
Before using AI each time, ask yourself these five questions. Tick each one to confirm you are ready:
🎮 挑戰你的理解 🎮 Test Your Understanding
你已閱讀完指引——接下來，透過互動小遊戲熟悉每一條規則。 You've read the guidelines—now reinforce each rule through interactive games.
進入遊戲挑戰 → Enter games →