Is Using AI to Write Your Essay Cheating? What the Rules Actually Say

"Did you use AI?" is the wrong question — nobody asks whether you used a spell-checker. The questions that decide it are what did you ask it to do, whose thinking is on the page, and did you say so. Here is what the written rules actually require.

The rules exist — they are just in two different places

Students look for one answer and find none, because AI rules live in two separate worlds: publishing ethics bodies, which have converged on a clear position, and university policy, which deliberately hasn't.

What publishers agree on

On one point there is no ambiguity anywhere. The Committee on Publication Ethics (COPE) states that AI tools cannot be listed as an author, because they cannot take responsibility for the submitted work and — not being legal entities — cannot declare conflicts of interest or agree to copyright and licence terms. The ICMJE says the same for medical journals: chatbots cannot be authors because they cannot be responsible for the accuracy, integrity and originality of the work.

Where they differ is only in the paperwork:

ICMJE — describe AI used for writing, editing or proofreading in the acknowledgements; report AI used to collect or analyse data or generate figures in the methods.
Elsevier — add a section headed "Declaration of Generative AI and AI-assisted technologies in the writing process" at the end of the manuscript, before the references.
Springer Nature — LLMs do not meet the authorship criteria, and their use must be documented in the Methods section or equivalent.

All three also carve out the same exemption: routine copy editing doesn't need declaring. Elsevier excludes "basic tools" for grammar, spelling and references; Springer Nature excludes "AI assisted copy editing" — improvements to human-written text for readability, style, grammar, punctuation and tone. Nobody is coming for your spell-checker.

What universities do instead

Universities have not written a single rule, and that is on purpose. The Russell Group's 2023 principles on generative AI in education commit its universities to supporting AI literacy, adapting teaching and assessment to incorporate ethical AI use, and upholding academic rigour and integrity — a framework for each institution to fill in, not a blanket ban.

In practice many departments now set the permitted level per assignment, often using something like the AI Assessment Scale (Perkins, Furze, Roe & MacVaugh, Journal of University Teaching & Learning Practice, 2024), which runs from no AI use at all through to full AI integration. Which means: the rule that governs you is the one on your assignment brief, not the one your friend on another course was given.

A green / amber / red guide

Assuming your brief doesn't say otherwise:

Green — usually fine. Fixing grammar and typos. Asking for search terms or synonyms to try in a database. Having a concept explained to you, then checking it against a real source. Getting feedback on the clarity of a paragraph you wrote.
Amber — allowed on many courses, declare it. Rewriting your own sentences for flow. Turning your notes into an outline. Summarising a paper you have actually read. Writing or debugging analysis code. The work is still yours; the assistance is worth naming.
Red — misconduct on almost any course. Submitting AI-generated argument or analysis as your own. Including citations you never opened. Presenting AI-generated data, quotes or figures as real. Using AI at all where the brief forbids it.

One test cuts through most amber cases: if a marker asked "why did you argue this rather than that?", could you answer from your own reading? If yes, the thinking is yours. If no, you are submitting someone else's work — and the someone else is a language model.

The thing that actually gets people caught

It usually isn't a detector. Detection is probabilistic and contested — worth understanding, but not the real exposure (see what Turnitin's AI detector can and can't tell you). The thing that ends conversations is a supervisor typing one of your references into Google Scholar and finding that the paper, the volume or the DOI does not exist. Language models generate plausible-looking references that were never published, and a fabricated source is a fabricated source whether or not you meant it.

So the non-negotiable habit is simple: open every source before you cite it. Confirm the authors, year, title and venue against the actual record — not against what the chatbot told you they were.

Verify first, format second

Once you have the real source open in front of you, our free citation generator turns it into a correct reference and in-text citation in APA, MLA, Chicago, Harvard or IEEE — so the only step left is the one that matters.

How to declare AI use, concretely

In an essay or dissertation, a sentence in the methods or introduction usually does it:

I used ChatGPT (OpenAI, 2025) to suggest alternative phrasings for the discussion section and to generate initial database search terms. All sources were located, read and verified by me; all analysis and argument are my own.

APA advises describing how you used the tool in your Method section — or, for an essay or literature review, in your introduction — and giving the prompt alongside any generated text you reproduce. For the reference-list entry, treat the model as software:

OpenAI. (2023). ChatGPT (Mar 14 version) [Large language model]. https://chat.openai.com/chat

For a journal submission, follow the target journal's wording. Elsevier supplies a template: "During the preparation of this work the author(s) used [NAME OF TOOL] in order to [REASON]. After using this tool/service, the author(s) reviewed and edited the content as needed and take(s) full responsibility for the content of the published article." That last clause is the whole principle in one line — accountability never transfers to the tool.

Three questions before you submit

Does my brief or module handbook say what level of AI use is permitted here — and have I actually read it?
Have I opened and read every source in my reference list?
If asked, could I defend every claim in this piece from my own reading?

Three yeses and you are not cheating — you are using a tool, which is what students have always done. The trouble starts when AI stops assisting your thinking and starts replacing it.

Write with AI you can actually declare

CiteGood drafts from real, verifiable sources and formats every citation for you — nothing invented, nothing to quietly hope nobody checks.

Try CiteGood free →

Frequently asked questions

Is using ChatGPT to write an essay cheating?

It depends on what you asked it to do and what your institution allows. Policies vary by university, department and even by module, and the rule that applies to you is the one on your assignment brief. Submitting AI-generated argument and analysis as your own work is generally misconduct; using AI to brainstorm, explain a concept or check grammar is usually permitted. When a brief is silent, ask your tutor before you submit — not after.

Can AI be listed as an author on a paper?

No. COPE's position statement on authorship and AI tools says AI cannot be listed as an author because it cannot take responsibility for the submitted work, and as a non-legal entity it cannot declare conflicts of interest or manage copyright and licence agreements. The ICMJE takes the same position for medical journals, and publishers including Elsevier and Springer Nature apply the same rule.

Do I have to declare using AI for spelling and grammar?

Usually not, in publishing. Elsevier's declaration requirement excludes basic tools used to check grammar, spelling and references, and Springer Nature does not require a declaration for 'AI assisted copy editing' — improvements to human-written text for readability, style, grammar, spelling, punctuation and tone. Coursework is a different matter: some universities want any AI use acknowledged, so check your own policy.

Where do I declare AI use in a manuscript?

It depends on the publisher. Elsevier asks for a section titled 'Declaration of Generative AI and AI-assisted technologies in the writing process' at the end of the manuscript, before the reference list. Springer Nature asks for LLM use to be documented in the Methods section or equivalent. The ICMJE asks for AI used in writing or editing to be described in the acknowledgements, and AI used to collect or analyse data or create figures to be reported in the methods.

How do I cite ChatGPT in APA?

Treat the model as software: OpenAI. (2023). ChatGPT (Mar 14 version) [Large language model]. https://chat.openai.com/chat. APA also advises describing how you used the tool in your Method section — or, for an essay or literature review, in your introduction — and including the prompt you used alongside the relevant generated text.