What Students Should Check Before Submitting AI-Assisted Work

AI tools are now normal in student workflows. The rules governing them are not settled, they differ sharply between institutions and even between modules, and the consequences of getting it wrong are serious.

Work through this before you submit anything you used AI on.

1. Read the actual policy

Not what a friend said, not what a forum said. Your institution's academic integrity policy, and your module handbook, which may be stricter.

Policies generally fall into a few categories:

• Prohibited entirely — any AI use is misconduct.
• Permitted for specified purposes — commonly brainstorming, outlining, grammar; not generating prose.
• Permitted with disclosure — allowed if you declare what you used and how.
• Unrestricted — rare, but it exists.

If it is unclear, ask your tutor in writing and keep the reply. An email saying "yes, that's fine" is worth a great deal if a question arises later.

2. Declare it if declaration is required

Where disclosure is expected, be specific and factual: which tool, at which stage, for what. "I used an AI assistant to generate an initial outline and to rewrite two paragraphs for concision" is a proper declaration.

Under-declaring is treated far more harshly than over-declaring. Nobody has ever been penalised for excessive honesty about their process.

3. Verify every citation

This is where AI-assisted work most often comes apart. Fabricated references are easy for a marker to spot — they check one, it does not exist, and now every other source is suspect.

Open every source. Confirm it exists, confirm the author and year, and confirm it says what you claim. The full method is in how to fact-check AI-generated writing.

4. Check whether it sounds like you

Markers who have read your previous work notice a change in voice. Generated prose has recognisable habits:

• Relentlessly even sentence length and rhythm.
• Formulaic transitions — *moreover*, *furthermore*, *in conclusion*, in that order.
• Confident generalities with no specific evidence attached.
• Balanced hedging on every point, so the essay never commits to anything.
• A vocabulary noticeably more elevated than your usual register.
• Tidy three-item lists everywhere.

Read it aloud. If it does not sound like you talking, rewrite it until it does. This also usually improves the essay.

5. Confirm you understand your own argument

A practical test: close the document and explain the argument out loud, in your own words, in two minutes.

If you cannot, you have a problem beyond academic integrity — you cannot defend it in a viva, a seminar, or an exam that assumes you learned this material. Submitting work you do not understand fails at the thing the assignment was for.

6. Look for the substance gaps

Generated text is fluent, which disguises thinness. Check specifically for:

• Paragraphs that restate the thesis without advancing it.
• Claims with no evidence attached.
• A counterargument section that does not actually engage with the objection.
• Assertions about the module's set texts that you have not verified against the texts.

7. Check it answers the question set

AI answers the prompt you gave it, which is not necessarily the question on the assignment brief. Reread the brief and confirm you have addressed what was asked — including any specific requirements about sources, theoretical framework or word count.

Where the line sits

The reasonable version of AI use in academic work looks like this: the ideas are yours, the argument is yours, the sources are real and read, and the words are ones you could have written and can defend.

The version that gets people into trouble is generating an essay, changing a few words, and submitting it. It is usually obvious, it is normally misconduct, and it means you learned nothing from the assignment.

AI Essay Writer Assistant includes rewriting and originality tools precisely because the editing pass matters. Use them on your own draft rather than as a way to disguise someone else's.