Find your own baseline score
A detection score means very little in isolation and quite a lot compared against your own writing. Half an hour of old coursework gives you the comparison.
The problem with a single score is that you have nothing to hold it against. Is 28% high? For whom, on what kind of writing?
The fix is cheap. Run three or four pieces you wrote before any of this was a question, and you have your own number. After that every score you see is a comparison rather than a verdict.
How to build one
Pick old work. Coursework, a report, a piece of writing from before generated text was something anybody worried about. Provenance is not in doubt, which is the point.
Pick a spread of types. One argumentative essay, one technical or methods-heavy piece, one piece of reflective or personal writing. They will score differently and the difference is the useful part.
Run each one whole, not an extract. Every detector is more reliable on more text.
Write down the document score and the highest paragraph score for each.
Thirty minutes, once.
What it tells you
Your typical range. If your old, unquestionable work sits at 20 to 30%, then 28% on a new piece is your normal writing rather than a signal.
Which registers push you up. Nearly everybody scores higher on technical and methods writing than on argumentative writing, because formulaic sections are formulaic by design. Knowing your own gap means you can read a high score on a methods section correctly.
Whether you are in the exposed group. Some people write in a naturally regular way: even sentence rhythm, conventional constructions, careful revision. Our corpus puts human academic writing at a within-document sentence-length standard deviation of 8.7 words and generated text at 6.4. If your own writing sits close to the lower figure across every sample, that is a property of your style, and it is worth knowing about yourself before somebody else discovers it.
Which habits are yours. Look at the paragraphs that score highest across all your samples. The same two or three things usually recur: connective openings, an even rhythm, a tendency to generalise where you could name the thing. Those are your habits, and they are fixable once named.
Using it afterwards
A new piece scoring inside your range is your normal writing. A new piece well above it is worth a look, and now you know which paragraphs to look at rather than guessing.
That converts the whole question from an anxious yes-or-no into an ordinary comparison against a number you own.
Why this needs an uncapped tool
Building a baseline means running several full documents and then re-running them as you learn what moves them. A word cap makes it impossible, and checking fragments gives you the least reliable version of every answer. See why detectors are unreliable on short text.
Ours is free and unlimited with no account, takes .pdf, .docx and .tex, has no word cap, scores every paragraph separately, and never stores your text. Wrong-flag rate 0.4%, which is 99.6% specificity, measured on 15,900 documents of real academic writing.