Resubmitting after an AI flag
If you get the chance to revise and resubmit, the changes that move a score are specific, measurable, and mostly not the ones people try first.
Whether a resubmission is available at all is set by your institution's own procedures, which vary a great deal and which we are not the right people to describe.
What we can describe is what changes a score and what does not, because that part is the same regardless of who is asking.
What does not move it
Synonym substitution. Swapping words for less common ones changes vocabulary without touching structure, and structure carries most of the signal. It also degrades the writing. See does QuillBot get flagged.
Another editing pass. This is the one people reach for first and it moves the score the wrong way. Every revision smooths prose, and smoothing removes the variation that reads as human. A carefully polished resubmission scores higher than the draft it replaced.
Adding errors deliberately. Detectors do not score correctness. Typos cost you a reader and buy nothing.
Unusual punctuation. Same reason. Not what is being measured.
What does move it
Three things, all structural, none of which change what you are saying.
Sentence length variation. The strongest single signal. Human academic writing in our corpus has a within-document standard deviation of 8.7 words, generated text 6.4. Find a run of three sentences of similar length, split one and merge another pair. A four-word sentence beside a thirty-word one does more than an hour of word substitution.
Connective openings. "Moreover", "Furthermore", "Additionally", "Overall". These open 2.4% of human sentences and 5.8% of generated ones. Most can be deleted outright, and if the logical relationship is real the sentence order already carries it.
Specific detail. Put back what you generalised. Name the method rather than "various approaches", give the number rather than "significant improvement". Specificity is the strongest human signal there is and it is the first thing lost when tightening a draft.
Full list in what makes text read as machine written.
Work paragraph by paragraph
A document score is an average and it hides where the work is. In most documents two or three paragraphs carry the signal and the rest are fine.
Score per paragraph, sort by which sits highest, and fix those. That is usually twenty minutes rather than a rewrite, and it leaves the rest of your document alone.
Pay particular attention to the sections that score high for structural reasons: methods and procedures are formulaic by design, and literature reviews are long, hedged and repetitive by necessity. See AI detection false positives, by type of writing.
Measure the before and after
The point of scoring per paragraph is that you can see the change rather than hope for it. Run the original, make the three changes above on the highest paragraphs, run it again.
Ours is free and unlimited with no account, takes .pdf, .docx and .tex, has no word cap so you can run the whole thing as many times as you like, and never stores your text. Our wrong-flag rate on genuine human writing is 0.4%, published, against Turnitin's 4%.