Is humanising a text cheating? Where the line runs
Humanisation is a rewrite that changes a text's structure and rhythm. In itself it is not cheating — an editor does the same job. It becomes cheating when its purpose is to hide that a model was involved somewhere you were supposed to declare it. What decides is the intent and the rules for your assignment, not the tool.
When it is ordinary work on a text
Rewriting a clumsy paragraph, shortening a sentence, making the tone consistent across a document — every newsroom does this and nobody objects.
The same goes for having your own text rewritten so it reads better. Authorship does not change; the content stays yours.
When it is dodging the rules
If the rule says declare your use of AI and you instead rewrite the text so it will not be noticed, you are breaking it regardless of the resulting score.
A low number in a detector legalises nothing. Rules talk about how a text came about, not about how it looks.
How to decide
Try a simple test: could you write down what you did in a note at the end of the work? If yes, you are fine. If not, you know why.
In DetekceGPT you also see the machine feel before and after a rewrite. That is a guide to the quality of the rewrite, not a confirmation that something is allowed.
FAQ
Can you tell a text went through humanisation?
A rewrite changes structure and rhythm, so the general score usually drops. A departure from your earlier texts can still hint at the change.
Is humanisation useful if I write myself?
Yes, as editorial help. For your own text, though, reading it aloud usually does the job.
What if the rules say nothing about AI?
Then the general principle that the work is your own applies. A short note about what you used costs nothing and ends the argument in advance.
Try it on your own text
Paste a text or upload a file and look at the score and at the specific sentences that came out suspicious.
Run a detection