Is rewriting my own writing with a tool cheating?

Under most policies, no. Editing prose you wrote is closer to what a grammar checker does than to what an essay mill does, and very few institutions prohibit it. The ones that do usually say so explicitly, which is why the module handbook is the document to read.

Is rewriting AI-generated text to avoid detection cheating?

Where AI-generated submissions are prohibited, yes — and the rewriting makes it worse rather than better, because it is evidence of an attempt to conceal. That is the case this page is least equivocal about.

Does disclosing it make it acceptable?

Only if the use was permitted in the first place. A declaration is a record, not a permission slip. Where a tool was banned, disclosing it afterwards is honest and still a breach; where it was allowed, the declaration is simply what the policy asked for.

What if my institution has no policy on this?

Then the institution-wide rules apply underneath the course, and if both are silent, ask in writing before you submit. An emailed answer costs you nothing and gives you a document; a decision made afterwards about what you assumed is much harder to defend.

Is it different for work outside education?

Usually, because there is no integrity policy — a client contract, a publisher's author guidelines or an employer's rules take its place, and several journals now require disclosure of AI assistance. The question stops being misconduct and becomes what you agreed to.

Do you have a commercial interest in this answer?

Obviously. We sell a rewriting tool, and a page telling you that two common uses of it are misconduct is against our short-term interest. It is here because the alternative is telling students what they want to hear about a decision that can end a degree.

Is using an AI humanizer cheating?

It depends on two things: the rule you are under, and what you put into the tool. Rewriting your own prose is permitted almost everywhere. Rewriting generated text to get it past a check is misconduct under almost any policy, and the rewriting is what makes it hard to explain.

Last reviewed 15 August 2026

The question is really two questions

People ask this as though the tool were the thing being judged. It is not. Every policy in this area judges what you submitted and how you produced it, so the same software sits on both sides of the line depending on what you fed it.

Take the two ends. You wrote a draft, you find it stiff, and you run it through a rewriter to loosen the rhythm — the ideas, the evidence and the argument are yours, and what changed is phrasing. Now the other end: a model wrote the essay, you run it through a rewriter, and you submit the result on a course that prohibits generated work. Both used the same tool. Only one of them is a question about phrasing.

Almost every real case sits somewhere between, which is why a general answer is useless and the specific rule matters so much.

Where the rule actually lives

Not on the university's public page. The rule that will be applied to your assignment is usually the course-level policy in the module handbook, which can be stricter than the institution-wide one and frequently is. Two courses in the same department can hold opposite positions in the same term and both are valid.

In US institutions an honour code may govern instead, and that changes the shape of the question rather than the answer: a code states a standard and expects you to reason from it, so “the rules did not mention rewriting tools” carries less weight than it would under an enumerated rulebook.

If both are silent, ask before submitting and keep the reply. That is a five-minute email that produces a document, and a document is worth considerably more than a recollection of what you assumed was fine.

The two uses that are misconduct nearly everywhere

Submitting generated work as your own where that is prohibited. The rewriting is irrelevant to whether this is a breach — it was a breach before the tool was opened. What the rewriting adds is intent. An investigator looking at generated content is considering carelessness or misunderstanding; an investigator looking at generated content that has been processed to read differently is considering concealment, and those attract different outcomes.

Declaring something untrue. If you are asked whether you used AI assistance and you say no, the false declaration becomes its own offence, usually a more serious one than the original use. It is also the part that is hardest to argue about later, because unlike a detection score it does not depend on anyone's statistics.

What a rewrite does not do

It does not discharge a disclosure obligation. If your policy requires you to declare AI assistance, running the output through another tool does not remove the assistance from the history of the document — it adds a second tool to the list.

It also does not deliver a detector outcome, and this site does not claim one. What we have said elsewhere about whether humanizers work applies exactly here: rewriting genuinely changes prose, and nobody can promise you a number on software they do not control. Buying a tool on that promise means buying a guess.

The uncomfortable part

We sell a rewriting tool, so a page telling you that two common uses of it are misconduct costs us something. It is here anyway, for a reason that is easy to state: the readers most likely to find this page are the ones with the most at stake, and a business that answers them with reassurance is trading their degree for a subscription.

The version we are comfortable selling is the first one at the top — your writing, improved. That is a real product with a real use, and it does not require anyone to be misled about what a policy says.

If you have already been accused

This page is about the decision before submission. If the question has already been raised about work you handed in, the priority order changes entirely and what to do when you are accused is the place to start, with the evidence that actually helps next — which is drafting history rather than anything a tool can produce afterwards.

Common questions

Is rewriting my own writing with a tool cheating?: Under most policies, no. Editing prose you wrote is closer to what a grammar checker does than to what an essay mill does, and very few institutions prohibit it. The ones that do usually say so explicitly, which is why the module handbook is the document to read.
Is rewriting AI-generated text to avoid detection cheating?: Where AI-generated submissions are prohibited, yes — and the rewriting makes it worse rather than better, because it is evidence of an attempt to conceal. That is the case this page is least equivocal about.
Does disclosing it make it acceptable?: Only if the use was permitted in the first place. A declaration is a record, not a permission slip. Where a tool was banned, disclosing it afterwards is honest and still a breach; where it was allowed, the declaration is simply what the policy asked for.
What if my institution has no policy on this?: Then the institution-wide rules apply underneath the course, and if both are silent, ask in writing before you submit. An emailed answer costs you nothing and gives you a document; a decision made afterwards about what you assumed is much harder to defend.
Is it different for work outside education?: Usually, because there is no integrity policy — a client contract, a publisher's author guidelines or an employer's rules take its place, and several journals now require disclosure of AI assistance. The question stops being misconduct and becomes what you agreed to.
Do you have a commercial interest in this answer?: Obviously. We sell a rewriting tool, and a page telling you that two common uses of it are misconduct is against our short-term interest. It is here because the alternative is telling students what they want to hear about a decision that can end a degree.