Most guides on writing essays with AI fall into two camps: the ones that show you how to paste and submit, and the ones that tell you never to touch the tools at all. Both are wrong. AI is genuinely useful for writing essays — for research, for structure, for drafting, for editing — and using it well is a skill worth learning. But the line between using it and letting it write for you is real, and crossing it is academic misconduct at almost every university.
This guide gives you a workflow that keeps the ideas yours at every step: outline, draft, critique, revise — with a disclosure template at the end.
First: what is allowed, and what is not
The rules vary by course, and the syllabus is the final word — check it before you start. But the pattern is consistent across European universities:
Usually allowed: brainstorming ideas, building an outline, explaining a concept, finding sources, checking your grammar, editing your own draft, summarising a paper you have read.
Usually misconduct: generating an essay and submitting it as your own, generating the argument and the evidence, paraphrasing AI output to hide that it is AI output, submitting AI-generated references you have not read.
The principle underneath is simple: the thinking must be yours. AI can help you think — it can challenge your argument, find a counterargument, explain a paper — but it cannot do the thinking for you. If you could not defend every sentence in your essay in a viva, you have gone too far.
Step 1: Research first, AI second
Start without AI. Read the question, read your sources, and write down your own answer in a few sentences — messy, incomplete, yours. Then use AI to test it: ask ChatGPT or Claude to find the strongest counterargument to your position, or to explain the one paper you did not understand. Our guides to ChatGPT for students and to Claude vs ChatGPT explain which tool fits which job.
Then find your sources with a tool that cites them. Perplexity answers with sources you can open and check; Gemini Notebook answers only from the sources you upload, with a citation to the exact passage. Never build an argument on a reference you have not opened. AI tools invent plausible-looking references — wrong authors, wrong years, journals that never published the paper — and a fake reference in your bibliography is the fastest way to fail an essay.
Step 2: Build the outline yourself
The outline is where the essay is won or lost, and it is the step AI is worst at doing for you. An AI outline is a generic outline: introduction, three body paragraphs, conclusion, with the same structure for every question. Your outline should carry your argument: what you claim, what evidence supports it, what the counterargument is, and why your claim still stands.
Write the outline yourself, in your own words, one line per paragraph. Then — and only then — show it to AI and ask: "What is missing from this outline? What is the strongest objection to it? Which paragraph is the weakest?" That is AI doing what it is good at: critiquing your thinking, not replacing it.
Step 3: Draft in your own voice
Write the first draft yourself. Not because AI cannot write it — it can, fluently — but because the draft is where you discover what you actually think. The sentence you struggle to write is the sentence that tells you your argument has a hole. If AI writes it for you, you never find the hole.
If you are stuck, use AI as a sparring partner, not a ghostwriter. Ask it to draft one paragraph from your outline — one — and then rewrite it in your own voice, keeping the structure and changing the sentences. Or ask it for three alternative openings and pick the one that sounds like you. The draft that comes out of that process is yours: your argument, your examples, your voice, with AI's help on the phrasing.
Step 4: Let AI critique, not rewrite
Once you have a full draft, AI becomes genuinely useful again — as a critic. Ask it to read your essay and answer specific questions: "Where does my argument weaken? Which paragraph would a sceptical lecturer attack? Is there a claim here that my sources do not support?" A good critique is worth more than a good draft, and it is the use of AI that no university objects to.
Then edit with tools built for editing. Grammarly Free catches spelling and grammar and gives you 100 AI prompts a month; its Pro plan ($12 a month billed annually, or $30 monthly) adds full-sentence rewrites, tone adjustment and a plagiarism check. QuillBot Free paraphrases up to 100 words at a time in two modes; Premium ($8.33 a month billed annually) removes the cap and adds a plagiarism checker covering 25,000 words a month.
A worked example: one paragraph, three ways
Here is what the workflow looks like on a single paragraph. Your outline says: "Sleep improves exam performance, but only for morning exams — the evidence is correlational, not causal."
Your first draft: "Sleep is good for exams. Students who sleep more do better. A study found this." It is true, but vague — it does not say which exams, how much sleep, or what the study actually found.
AI's critique: you ask ChatGPT to attack the paragraph, and it points out three holes: no numbers, no limits, no source. It asks: how many hours? Which students? Was the effect causal?
Your revision: "In a 2023 study of 200 first-year students, those sleeping 7–9 hours scored higher on morning exams than those sleeping under 6 hours. The effect was correlational — the authors warn that sleep may mark, rather than cause, better preparation — and it appeared only for morning exams." Same argument, now defensible, with numbers and limits. The AI did not write it. It showed you what was missing, and you wrote it.
That is the whole workflow in one paragraph: your idea, AI's critique, your revision. Repeat it for every paragraph and you have an essay.
What to do when you are stuck
Stuck usually means one of three things, and each has a different fix. If you do not know what to say, you have not read enough — go back to the sources, not to the chatbot. If you know what to say but not how to structure it, write the outline as bullet points and let AI challenge it. If you have the argument but the sentences will not come, write the ugly version first — short, plain, even wrong — and edit it later. A bad draft you can fix beats a blank page you cannot. The chatbot is most useful for the second problem and least useful for the first, and knowing which problem you have is half the skill.
The paraphrasing trap
Here is the mistake that gets students caught. You generate a paragraph with AI, run it through QuillBot to reword it, and submit it as your own. That is not editing — it is paraphrasing to evade detection, and universities treat it as plagiarism. Turnitin names QuillBot as a paraphrasing tool its systems may flag, and the reworded text still carries the AI's argument, its structure and its evidence. Changing the words does not make the ideas yours.
The rule is simple: paraphrasing tools are for improving your own writing, not for disguising someone else's — and AI output counts as someone else's. If a sentence started in an AI tool, either rewrite it yourself from your own understanding, or do not use it. A paraphraser cannot turn AI's work into yours.
Step 5: Check every reference, then disclose
Before you submit, verify every reference in your bibliography against the real source. Open each one. Confirm the author, the year, the journal, and — most importantly — that the page says what your essay claims it says. AI-generated references are wrong often enough that this check is not optional. Our guide on how to cite AI tools gives you the current APA, MLA and Chicago formats, and the disclosure statement EU universities want.
Then disclose your AI use the way your university asks. Many European universities now require a statement in the report — Utrecht's Graduate School of Life Sciences, for example, requires a Statement of Scientific Integrity in the final report even if you did not use AI at all, and a complete prompt log if you did. KU Leuven's rule is transparency: hidden AI use is an irregularity under Article 84 of its education regulations. Keep your drafts and your prompt log as evidence of your own work — they are what prove the essay is yours if anyone asks.
Here is a disclosure template you can adapt (check your university's exact wording first):
"This essay used generative AI for the following: brainstorming the outline, finding sources, and editing grammar in the final draft. All arguments, analysis and conclusions are my own. AI-generated text was not submitted as my own work. A log of the prompts used is available on request."
Keep your evidence
Save three things for every essay: your drafts (so you can show the essay developing), your prompt log (so you can show what you asked the AI and what it returned), and your sources (so you can show where every claim came from). If a lecturer ever questions your work, that folder is your defence — and it is also, incidentally, the best study material you will ever make for the exam on the same topic.
Frequently asked questions
Can I use ChatGPT to write my essay?
You can use it to research, plan, draft and edit. Submitting its output as your own work is academic misconduct at almost every university. Use it to think, write the essay yourself, and disclose what you used.
Will my university detect AI writing?
Maybe, and that is the wrong question. AI detectors are unreliable — Grammarly's own detector page says no AI detector is 100% accurate — and a false positive can happen to a human-written essay. The question is not whether you can get away with it, but whether the essay is yours. If it is, detection does not matter.
Is using Grammarly or QuillBot cheating?
No, for editing your own writing. Grammarly's grammar check and QuillBot's paraphrasing of your own sentences are normal editing tools. Using them to disguise AI-generated text is a different thing, and that is what gets flagged.
What if my syllabus says nothing about AI?
Ask your lecturer. "Nothing said" is not permission, and the default at most universities is that undisclosed AI use in assessed work is misconduct. A two-minute question before you start is cheaper than a disciplinary hearing after you submit.
The bottom line
AI can make you a better essay writer — if you use it to think, not to avoid thinking. Research first, build the outline yourself, draft in your own voice, let AI critique the draft, check every reference, and disclose what you used. The essay that comes out of that workflow is yours in the only sense that matters: you can defend every sentence in it. And that is the skill the essay is testing in the first place.
Related guides
- How to Cite AI Tools in APA, MLA and Chicago — Plus the Disclosure Statement EU Universities Want
- Claude vs ChatGPT for Essay Writing: Which Writes Better?
- ChatGPT for Students: How to Use It for Studying Without Cheating
- EU University AI Rules: What 5 Universities Actually Allow
Rules and prices checked against university policy pages and vendor pricing pages in October–November 2026. Your course syllabus is the final word — check it before you start. This article is for general information and is not legal or academic advice.

