Disclosure

Disclosure means saying when and how you used AI in a piece of work. It sounds like a confession. It is closer to a receipt. Done well, it takes about four sentences, and it is quickly becoming a normal professional habit rather than an admission of anything.

Why disclosure protects you

Imagine two students who used AI in exactly the same legitimate way. One wrote a short note saying so. One stayed quiet. If a question ever comes up, the first student has evidence of good faith, written before anyone asked. The second student has a secret, and secrets look like guilt even when there is none. Disclosure converts a risk into a record. It also does something subtler: knowing you will disclose changes how you use the tool, because you only do things you would be comfortable describing.

Timing is what gives the note its power. A disclosure written before submission is evidence, because you created it when you had nothing to defend, and a record made at the time cannot be reshaped by whatever happens later. The same words offered after a question is raised are just a story, and stories told under pressure are worth less however true they are. This is the reason receipts beat memories in any dispute: not because the person remembering is dishonest, but because the receipt existed before the disagreement did.

The fear that markers punish honesty

The most common reason students stay silent is not a plan to deceive anyone. It is the fear that a marker who sees an AI declaration will quietly think less of the work. That fear deserves to be taken seriously, because it feels reasonable and plenty of students share it, and then it deserves to be examined, because look at what it actually asks you to bet. Where a subject permits declared use, the worst case of declaring is a sceptical marker reading your genuinely own work more carefully, which honest work survives. The worst case of concealing, where declaration is required, is a misconduct finding, and that bet has to keep winning every time anyone ever asks about the work. One of those bets is settled at marking and never reopens. The other never finishes. The case study at the end of this topic follows one student on each side of exactly this choice.

What a good disclosure note looks like

• The tool: what kind of AI you used. The category matters more than the brand.
• The purpose: exactly what you used it for. Brainstorming, feedback on a draft, checking reasoning, generating practice questions.
• The boundary: what remained yours. The research, the analysis, the final words, the decisions.

For example: I used an AI chatbot to suggest possible structures for this report and to critique two drafts. All research, analysis and final wording are my own, and I checked every factual claim against my sources. That is the whole thing. Specific, short, and impossible to mistake for hiding something.

A weak note, and how to strengthen it

Now compare that with the note a rushed student writes: I used AI a bit to help with this assignment. Every word of it may be true, and it still fails at the one job a disclosure has, because the reader cannot tell what the tool did and what you did. A bit could mean fixing a comma or writing the conclusion. Help could mean explaining a concept or producing the entire analysis. Vague disclosure creates the very suspicion it was meant to prevent, so the fix is always the same: name the category of tool, list the specific uses, and draw the boundary of your own work in one plain sentence. And if you find the boundary genuinely hard to describe, that is not a writing problem. It is a sign the boundary was never clear while you worked, which is worth discovering before you submit rather than after.

Keep the trail as you go

Disclosure is easiest when there is nothing to reconstruct. If you use AI on a piece of work, keep the trail while it happens: save or link the chats, keep your early drafts, and let the version history in your document quietly do its job. A document that grew over two weeks, with your edits layered on top of one another, is very hard to fake and very easy to point to. The note then takes two minutes, because you are describing a record rather than recalling one, and if a question ever comes you hold something better than an explanation. You hold evidence.

The norms being written right now

You are learning this at the exact moment the adult world is working it out too. Scientific journals now require authors to declare AI use in papers. Universities are adding declaration sections to cover sheets. Publishers are setting rules for AI in books, and employers are writing policies for AI at work. The details differ, but the direction is the same everywhere: use is increasingly fine, concealment increasingly is not. Learning to disclose well now is learning a skill your future workplaces will expect.

Disclosure closes the loop the last lesson opened. Academic integrity is about making sure the claim you submit is true, and disclosure is how you put the true claim in writing. The next lesson turns the same instinct outward, because once you can describe your own AI use precisely, you are well placed to notice when a company describing its AI is being rather less careful.