Generative AI is now part of how most students work, and university policy has moved from blanket prohibition towards conditional, module-specific permission. That shift makes the rules harder to follow, not easier — because the answer to "am I allowed to use this?" now depends on which assignment you are writing.
Read the Policy That Applies to This Assignment
Institutions increasingly publish a tiered scale. A common pattern:
- No AI permitted — typically exams and skills assessments.
- AI permitted for preparation only — brainstorming, understanding concepts, planning. Not for generating submitted text.
- AI permitted with declaration — you may use it, and must state how.
- AI integral to the task — you are being assessed partly on prompting and critical evaluation of outputs.
The tier is usually stated in the assignment brief. Where it is not, ask your module leader in writing and keep the reply.
Uses That Are Almost Always Acceptable
- Explaining a concept you have not understood from the reading.
- Generating practice questions to test your own knowledge.
- Checking grammar and clarity of text you wrote.
- Suggesting a structure for you to then evaluate and adapt.
Uses That Are Usually Breaches
- Generating passages of the submission and presenting them as your writing.
- Producing a literature review, or citations you have not verified.
- Paraphrasing a source through a tool to evade similarity detection.
- Fabricating or "improving" data.
Fabricated References Are the Common Trap
Language models still invent plausible-looking sources — real authors, real journals, article titles that do not exist. Submitting one is a serious matter because it is unverifiable at source. Check every reference in the actual database or publisher site before it enters your list. If you cannot find it, remove it.
How to Declare AI Use
Where declaration is required, be specific and factual. For example: "ChatGPT (GPT-4, OpenAI) was used on 3 September 2026 to generate an initial outline for Section 3 and to check the grammar of the discussion section. All analysis, argument and source selection are my own. No generated text has been submitted verbatim." Some institutions require prompt logs in an appendix — check before you submit.
Interpreting AI Detection Scores Carefully
Detection tools produce a probability estimate, not proof. Two things are worth understanding:
- False positives happen, and research has repeatedly shown they fall disproportionately on writers using English as an additional language, and on formal, structured prose — which is exactly what academic writing is.
- A score is a prompt for a conversation, not a verdict. Most institutions require corroborating evidence before any allegation proceeds.
The practical defence is process evidence: keep drafts, notes, version history and reading annotations. A student who can show how the work developed is in a strong position regardless of what a percentage says.
Our Position
AssignmentToppers provides tutoring, structural feedback, referencing support and originality review. We do not produce work for submission as a student's own, and we encourage every student to check their institution's AI policy for each individual assignment. Using support ethically is your responsibility, and it is the only approach that protects your qualification.