How to Write an AI-Use Disclosure Statement for Your Assignment (2026)

For a few years, the loudest question in academia was whether students were using AI at all. That question is settled. Recent surveys put generative AI use among students well above ninety percent, so banning it outright has become close to unenforceable. The rules have shifted accordingly.

The dominant policy across universities in 2026 is no longer prohibition. It is transparency. You are increasingly allowed to use AI for permitted tasks, so long as you say so clearly and honestly. Institutions like Princeton, Oxford, Columbia, Imperial, and Harvard’s education school now expect a disclosure, and many virtual learning environments add an AI-use checkbox right next to the old plagiarism declaration.

That is good news, because a clear disclosure protects you. The risk is no longer “I used a tool.” The risk is being caught having used one without saying so. This guide walks through what a disclosure statement is, what to put in it, where it goes, and how to word it so an examiner reads it as honesty rather than a confession.

What an AI-Use Disclosure Statement Actually Is

A disclosure statement, sometimes called an AI acknowledgement or an attribution statement, is a short, honest note that tells your marker three things: which AI tool you used, what you used it for, and how the output made it into your work. It sits alongside your integrity pledge, not in place of it.

It is worth being clear about what a disclosure is not. It is not an admission of cheating. It is not a request for permission after the fact. And it does not turn unpermitted use into permitted use. If your module says no AI for a given task, disclosing it does not make it allowed. Disclosure is how you stay honest about the help you were actually allowed to take.

Think of it the way you already think about citing a source. When you lean on someone else’s work, you name it. AI is the same idea, extended to a tool.

Step 1: Check What Your Course Actually Permits

Before you write a single line of disclosure, read your specific module guide or syllabus. This is the step students skip, and it is the one that matters most, because AI rules in 2026 are set at the course level, not the university level. One lecturer may welcome AI for brainstorming and outlining. Another may forbid it entirely for a particular essay. A third may allow it only for grammar and proofreading.

Look for three things in the policy:

• Whether AI is permitted at all for this assessment, and for which tasks specifically.
• Whether you need to disclose, and in what format the department wants it.
• Whether you must attach evidence, such as your prompts or a copy of the AI output.

If the policy is unclear, ask your lecturer in writing before you submit and keep the reply. A short email saying “I used a tool to check grammar, is that within the rules for this essay?” is a five-minute insurance policy. When you are unsure whether AI even belongs in academic work, our honest breakdown of what AI can and cannot do for a literature review is a useful reality check before you start.

Step 2: Know What Belongs in the Statement

A good disclosure is specific. Vague statements like “I used AI to help” tend to raise more questions than they answer, because a marker cannot tell whether you mean spell-check or a ghost-written paragraph. Aim to name each of these clearly:

• The tool and its version. “ChatGPT (GPT-4o)” or “Claude Sonnet” tells the reader far more than “an AI tool.”
• The task. Brainstorming search terms, outlining structure, checking grammar, summarising a paper you had already read, and so on.
• The scope. Which parts of the work the tool touched, and which it did not.
• The date, and where relevant, the prompts you used.
• What you did with the output. This is the important one. “I rewrote the suggestions in my own words and checked every claim against my sources” reads very differently from silence.

The more precise you are, the more the statement works in your favour. Specificity signals that you were in control of the process, which is exactly what process-based assessment is looking for. On that shift, see our piece on why AI detection is being replaced by process-based evaluation.

Step 3: Put It in the Right Place

Most institutions accept a disclosure in one of three spots, so use whichever your course specifies, and default to the last one if it is silent:

• In the body of the work, usually a short note under the title or in a footnote for a light use.
• In the acknowledgements, a natural home for a sentence or two.
• In an appendix at the end, which is the safest choice when you have prompts, screenshots, or a longer explanation to include.

If your tool use rises to the level of a source you drew content from, and not just a process aid, you may also need a formal reference entry. Keep the acknowledgement statement and the reference entry separate. The acknowledgement explains your process in plain language. The reference follows your citation style. Our six citation styles compared guide covers where these entries live in each system.

Step 4: Format the Reference (APA Example)

If your department uses APA and asks you to cite the tool as a source, the 2026 approach treats the company as the author. The pattern is straightforward:

A worked example looks like this:

Two practical notes. First, some AI tools let you share a link to the exact conversation, which is worth including when your department wants evidence. Second, the rules for citing AI are still settling, and different faculties word them differently, so treat your own style guide as the authority. If you are working in APA generally, our full APA citation format guide covers the surrounding reference list.

Copy-and-Adapt Disclosure Templates

Here are three honest templates for common situations. Adapt them to what you actually did, and never claim a smaller use than the real one.

For light use, such as grammar and structure:

AI acknowledgement: I used ChatGPT (GPT-4o) on 2 July 2026 to check grammar and suggest clearer sentence structure in my draft. I reviewed every suggestion, rewrote them in my own words, and take full responsibility for the final text.

For brainstorming and outlining:

AI acknowledgement: I used an AI assistant on 1 July 2026 to brainstorm search terms and to sketch a possible outline for this paper. The reading, analysis, and writing are my own. I checked every source myself and the argument reflects my own judgement.

For summarising papers you had already read:

AI acknowledgement: I used an AI tool to produce first-pass summaries of three articles I had already read in full. I verified each summary against the original source, corrected errors, and wrote the synthesis and conclusions myself. Prompts and outputs are attached in Appendix A.

Notice what every version has in common. Each one names the tool, the task, and the fact that you checked the work and own the result. That last clause is not decoration. It is the point.

Common Disclosure Mistakes to Avoid

A few patterns tend to cause trouble, and all are easy to sidestep:

• Being vague. “AI was used in this assignment” invites the reader to imagine the worst. Say what you actually did.
• Under-declaring. Describing a heavy use as a light one is worse than the original use, because it reads as an attempt to hide something.
• Disclosing unpermitted use as if a note fixes it. If the task forbade AI, a disclosure does not authorise it. Check first.
• Forgetting the evidence. If your policy asks for prompts or outputs, a statement without them is incomplete.
• Copying AI text unchecked. The disclosure is not a substitute for doing the work. You still have to read, verify, and rewrite.

That last point deserves weight. AI tools can produce fluent text that is confidently wrong, including invented citations that do not exist. A disclosure that says you checked everything only holds up if you actually did. Verifying every claim against a real source is the non-negotiable part of using any tool responsibly.

Where Litrevu Fits, Honestly

People sometimes assume a tool like Litrevu is the thing you would be nervous to disclose. It is built for the uses many university policies allow, often with a declaration: reading, searching and summarising your own sources.

Litrevu works from the papers you have already gathered and uploaded, and cites only those papers. You ask what they say, compare their methods and findings side by side, and check a paragraph you wrote against them, and every answer opens to the exact passage it came from so you can read it yourself. It can also draft cited text from your papers, but treat that as notes to think with, not text to submit: many universities do not allow AI-written text in submitted work, even after you edit it. Disclosing your use is then simple and honest: name Litrevu, say what you asked it to help you read and check, and confirm that the writing is your own.

Ready to try it on your own reading? Add the papers you already gathered, ask what they say, and open each answer’s passage to check it. Free to start, no credit card required. Try it free.