The safe rule is: use AI to understand and plan, but write and reference the work yourself. Submitting AI-generated text as your own, or paraphrasing it to hide where it came from, is what gets treated as misconduct.
Every university words its policy slightly differently and rules can vary between modules — so read yours in the first week rather than guessing at it in week nine.
Why first-years get caught out
Almost never by cheating. Overwhelmingly by guessing wrong.
Nobody hands you a clear rulebook in Freshers' Week. The policy exists, but it's a PDF three clicks deep in a VLE, written for staff, and it may be contradicted by whatever your seminar tutor said offhand in week two. Meanwhile 94% of UK undergraduates use generative AI for assessed work in some form, so “everyone does it” is true and tells you nothing about what your department permits.
Rules vary by module, not just by university. A history essay and a computing coursework on the same campus can have opposite policies on AI assistance, and the one that applies is the one printed on that assignment brief.
Three things to do in your first fortnight
- Find your academic-integrity policy. It's usually in the module handbook or on the VLE. Read the AI section — it's normally two paragraphs. Note whether your course requires you to declare AI use, because a surprising number now do, and failing to declare permitted use is itself a breach.
- Learn to reference properly, now. Unglamorous, and worth more marks than anything AI-related. Bad referencing sinks more first-year grades than misconduct ever will. Find out whether your course wants Harvard, APA or something departmental, and learn that one.
- Save your drafts as you go. Write in one continuously-edited document rather than assembling a clean copy at the end. If you're ever asked how you wrote something, the version history answers it for you in about four seconds.
What actually counts, in plain terms
- Fine on most courses: having a concept explained to you, generating angles to consider, checking a structure, correcting your grammar and spelling — then writing it yourself.
- Misconduct nearly everywhere: submitting generated paragraphs as your own words; running AI text through a paraphraser to disguise it; citing sources you never read, or that don't exist.
- Ask first: AI-summarised sources you then cite, translation from another language, and heavy redrafting of a generated first draft.
That last category is where the fabricated-reference problem bites hardest. A model will produce a plausible-looking citation with a real author, a real journal and an article that was never written. Submitting it isn't just an AI issue — it's a claim you read something you didn't, which is a more serious allegation than the AI use.
The habit that protects you for three years
Get these in during Freshers and you'll never have the 2am panic in third year: one document per essay from plan to submission, a reference saved the moment you use a source, and a policy you've actually read. That's it. It's not much, and it's the difference between a stressful degree and a calm one.
SafeGrade checks referencing, flags AI-looking writing and gives you a rough grade band before you submit — and Sage can tell you what your specific course allows.
Run one free scan →No card. No catch. First scan free.