Student guide, version 3.0.1, in force since September 2026
What am I allowed to do?
Every programme, every campus. What you may do with AI, what you must declare, what you risk and what you are entitled to. The actual rule, not a summary of it.
This guide completes the common core of the charter. Where the two diverge, the more protective provision applies. Read the common core
The essentials, in four lines
- AI assists you. You remain accountable.
- You declare your use of AI. Every time.
- You protect sensitive data.
- You check everything AI produces.
These four lines are enough for 95% of situations. ISC Paris AI Use Charter, student guide V3, section 1.
What you may do, what you must not
Allowed
Subject to the authorisation level stated for your assessment and to the five principles.
- BrainstormExplore a topic, ask for ideas, identify angles of analysis.
- StructureGet an outline, organise your thinking, rank your arguments.
- RewordImprove the clarity of your text, adjust the register.
- TranslateConvert your content into another language, then reread and validate it.
- ReviseCreate quizzes, flashcards and summaries to prepare for exams.
- CodeGet help on an exercise, debug, understand a piece of syntax.
- RehearsePractise for an oral, simulate a job interview, play out a case.
- UnderstandHave a difficult concept explained, ask for an example, get it in plain words.
Prohibited
- Have your work done for youHanding an assignment, a dissertation or a thesis entirely to AI.
- Hide your use of AILeaving out the statement, misrepresenting the assisted share.
- CheatUsing AI during an exam or an assessment where it is not permitted, N0 or N1 depending on the case.
- Pass on sensitive dataTo an unapproved AI: confidential information from your placement or apprenticeship company, other people’s personal data, content covered by professional secrecy.
- Copy and paste without checkingAI invents references, figures, quotations, sometimes entire pieces of legislation.
- Share your accessYour AI credentials, ISC or personal premium, are strictly personal.
- Falsify the statementLying in the AI use statement makes the situation worse, it does not settle it.
One use apart: calibrating
Knowing when to use AI and when to do without is a skill in itself. Researchers call it mode-switching: consciously moving between assisted and unassisted work. Before you send a prompt, ask yourself one question.
Mode-switching competence: ESCP Summit AI in Higher Education, March 2026 (Lee). Cited by the student guide V3.
What learning do I lose if I let AI do this?
A lot
Understanding a concept, building an argument, memorising. Do it yourself first. You can then ask AI to challenge your work.
Nothing
Formatting, technical translation, transcription. AI is your ally.
The five levels, on your assessments
| Level | Meaning | What you may do in practice |
|---|---|---|
| N0 | No AI | No recourse to AI. Invigilated exams, oral defences, work of personal expression. |
| N1 | Preparation | Brainstorm, research and organise with AI; write on your own. AI does not appear in what you submit. |
| N2 | Assistance | Reword, structure and translate with AI. AI use statement mandatory. |
| N3 | Collaboration | Use AI freely, with critical analysis of its output. Detailed statement. |
| N4 | Exploration | Demonstrate your critical command of AI itself. AI is the object of the assessment. |
By default
- Work submitted from home: level N2.
- Invigilated exams: level N0 or N1.
- Special cases: stated on the brief by the teacher.
The brief says nothing. The exact move
- Apply the default level above.
- Ask the teacher to confirm in writing, by email or Moodle message.
- Keep a record of the answer.
- Still in doubt? Ask before you start, never after.
A question asked always beats a breach recorded.
Declaring your use
This dissertation was produced with the assistance of Claude (Anthropic) for the structure of the outline and the rewording of the introduction and conclusion. The analysis is entirely my own. The sources cited were checked in Google Scholar and in the Cairn database of the ISC Paris library. Sector figures were cross-checked against the original sources (INSEE, Eurostat).
I used AI.
Far too vague. That is not a statement, it is an alibi: no tool, no parts concerned, no nature of assistance, no checks carried out.
Citing AI without granting it authority
An AI is not a primary source. Citing one to back a claim is a methodological error.
Citing AI in the bibliography does not exempt you from citing the original sources, articles, reports and data that you actually used and checked.
Template based on the French Ministry of Higher Education recommendations, DemoES framework, September 2025.
Citation template
[Tool and version] ([Publisher]), "[summary of the main prompt or objective]", type of use: [one-off / iterative], changes made: [summary], date accessed: [DD/MM/YYYY].
One-off use
ChatGPT-4o (OpenAI), "Rewording of the opening paragraph of chapter 2", type of use: one-off, changes made: rewording accepted at 70%, structure kept, examples added by hand, date accessed: 12/10/2026.
Iterative use
Claude Sonnet 4 (Anthropic), "Help structuring the dissertation outline and challenging the hypotheses", type of use: iterative (12 prompts over 3 days), changes made: outline revised twice, hypotheses 3 and 5 reworded following AI objections, date accessed: 05-08/11/2026.
Checking before you submit
- FactsConfirm every fact against an independent source.Where to lookOfficial sources, quality press.
- ReferencesType the exact title of every reference into Google Scholar.Where to lookIf it does not come up, it probably does not exist.
- FiguresCross-check every figure against a primary source.Where to lookINSEE, Eurostat, annual reports, original studies.
- ConsistencyReread the whole thing. Hunt for internal contradictions.Where to lookLogic, common sense.
- RelevanceDoes the content really answer the question asked?Where to lookGo back to the brief.
Reliable sources
- Google Scholar for academic references.
- ISC Paris library databases: Cairn, Business Source.
- Official sites: INSEE, Eurostat, governments, international organisations (OECD, IMF, WHO).
- Primary sources: annual reports, original studies, official accounts.
Sources to avoid on their own
- Wikipedia as the only source: a useful starting point, never a finishing one.
- Forums and social networks.
- Sites with no named author, no date, no publisher.
- AI itself as a source of authority. It is an assistant, not a primary source.
Placement, apprenticeship, group work
Between your company’s policy and the school’s, the more restrictive one wins.
- The company prohibits AIYou do not use AI for company work. Full stop.
- The company permits AIYou follow its rules and the ISC charter.
- The company has no written ruleYou ask your company supervisor before using AI. And you keep a record of the answer.
Junior enterprise and student-run companies
- You protect the data of your clients and partners.
- You declare AI use in your professional deliverables: the junior enterprise remains accountable for what it delivers.
- You only use tools whose terms of use and data retention you actually understand.
Group work
- The authorisation level applies to every member of the group, without exception.
- Each member must be able to account for their own contribution.
- The statement is collective and broken down by contribution: if Alice used Claude to reword part 2 and Bob used ChatGPT to translate part 3, both appear in it.
- If one member cheats, the whole group may be affected. Choose your teammates, and say no when one of them offers a shortcut.
Your rights, the sanctions, the appeal
Your rights
To be heard
The student is heard before any decision is taken. Always.
To human analysis
Human analysis prevails over automated AI detectors. No sanction is issued on the basis of a detector score alone: multiple indications, never a detector on its own.
To put things right
Flagging an omission quickly, before marking, is always received better than letting it surface late.
To appeal a decision
You may refer the matter to the director of studies within 15 days. The procedure is set out in the academic regulations.
The graduated principle
In the first months after the charter comes into force, the priority is education. That does not exempt anyone from the rules.
| Breach | Standard sanction | Procedure |
|---|---|---|
| First missing statement | Reminder of the rules | Educational discussion with the teacher |
| Repeated undeclared use | Zero for the work concerned | Written notification, copied to the director of studies |
| Authorisation level not respected | Zero for the work concerned | Written notification, copied to the director of studies |
| Established fraud: full delegation, falsified statement | Disciplinary board | Adversarial procedure, rights of defence |
| Repeat offence or serious case | Sanctions under the academic regulations, up to exclusion | Disciplinary board |
The charter creates no disciplinary regime of its own. These sanctions and their procedures come from the academic regulations, which remain the applicable text.
Challenging a level you find absurd
- Talk to the teacher concerned first. Make your case: why the level seems ill-suited, what you would propose instead.
- If the disagreement stands, refer the matter to the director of studies in writing, setting out your arguments. You receive a reasoned answer.
- You may also raise it with the extended AI Transformation Office, aito@iscparis.com, to flag a case worth feeding into the annual review. An elected student sits on it.
Challenging is legitimate. Ignoring the level while you do so is not.
AI is not a confidant
Conversational AI models are built to answer pleasantly, with the appearance of empathy, around the clock. That can create the illusion of a friendly presence or of psychological support. It is neither.
What AI is not
- It is not a human being. It feels nothing.
- It does not judge you, but it does not understand you either. It imitates understanding.
- It has no professional duty comparable to that of a doctor, a psychologist or a chaplain. It is bound by no professional secrecy.
- It can produce mistaken, sometimes dangerous medical or psychological advice, in a reassuring tone that makes the risk worse.
Risks to know about
- Emotional dependenceSystematically turning to an AI to talk about what weighs on you can deepen withdrawal rather than help you out of it.
- MisinformationA serious symptom misread, unsuitable advice in a moment of distress: the consequences are real.
- IsolationThe more you talk to a machine, the less you talk to people. It is mechanical.
If you are going through a hard time
You are not alone. Several people, trained and bound by confidentiality, are available to you.
3114
French national suicide prevention line
Free, confidential, 24 hours a day, 7 days a week.
- ISC Paris university health service (SSU)Reception, listening, referral. Contact details on the intranet, Student wellbeing page.
- Campus chaplainciesOpen to everyone, believers or not, no registration and no conditions.
- Your own doctor, or a psychologist in private practiceCover through student health insurance is possible.
- Student wellbeing intranet pageDirectory of internal and external resources.
AI can help you reword a letter, understand a concept in psychology, organise your thoughts before an appointment. It cannot hold your hand. For that, you need people.
Frequent questions
May I use the free version of ChatGPT for my assignments?
Yes, if the authorisation level allows it, N1 to N4, provided you declare your use and pass on no sensitive data. For ISC internal or placement data, use the approved tools.
AI produced false information in my assignment. Am I accountable?
Yes, entirely. That is the principle of Responsibility. You check everything AI produces before you submit it.
May I use AI to prepare for a placement interview or an oral?
Yes, and it is strongly encouraged. AI is excellent at simulating a panel, playing a recruiter, asking you unexpected questions. Practise.
My teacher says N0 out loud, but the written brief does not mention it. What do I do?
The teacher’s instruction prevails. Ask for written confirmation, by email or Moodle message, keep it, and apply N0.
May I record lectures to transcribe them with AI?
Only with the explicit consent of the teacher and of the students present. Without consent it is unlawful: GDPR, image rights, the teacher’s intellectual property.
A classmate lent me their premium AI account. Is that serious?
Yes. It breaches the tool’s terms of use and this charter. Use your own account or the approved ISC tools.
I used AI without declaring it. What now?
Contact your teacher immediately, before marking. Late honesty is always better than discovery. You will be heard.
An AI detector says my text was generated, but I wrote it myself. What do I do?
Do not panic. Detectors produce a great many false positives, especially on well-structured writing. You will be able to explain, and no sanction is issued on the basis of a detector score alone. Keep your drafts, your writing history, your notes: those are your evidence.
I feel I am becoming dependent on AI for my assignments. What can I do?
You have already done the hard part: noticing. Before each working session, write by hand what you are trying to understand, not what you are trying to produce. Set yourself AI-off stretches on learning tasks: reading, note-taking, first draft. Keep AI for formatting, challenge and proofreading. If the worry persists, talk to your director of studies or to the university health service. It is not a shameful subject, it is a generational one.
Where to ask, and whom
Two tools that answer straight away, one address for what the charter does not settle, three sessions to go further.
The two tools built for you
The AI Transformation Office
- Answer your questions on acceptable use.
- Point you to the right training.
- Clarify a situation you and your teacher cannot settle between you.
- Receive your incident reports, your suggested changes, your challenges to a level.
Training
- Studying with AI45 minThree twelve-minute micro-modules: what AI can and cannot do, what to expect from it in your own work, and where what you are allowed to do stops.
- The AI Act: five costly misconceptions45 minOpenly accessible. The European regulation, without the text of the regulation.
Online, at your own pace. Studying with AI awards a personal certificate from 75 per cent on the final quiz: dated proof that you took it.
A case this guide does not settle?
Write to aito@iscparis.com. The charter does not guess on your behalf, and neither does this page: where the text says nothing, the answer is asked for.
aito@iscparis.com