How accurate is our AI Check
We test our AI Check against real Turnitin results, on papers it has never seen.
98 of 100
papers we call Ready to submit also scored below 20% on Turnitin
83 of 100
sentences Turnitin flags are highlighted blue in our report
95 of 100
with purple added
3,166 papers with real Turnitin results, none used to train our model. Measured October 2026.
Your verdict
- Ready to submit — Submit as is. Our guarantee covers it.
- Almost ready to submit — A few sentences may be flagged. Humanize the colored ones.
- Humanize before you submit — Turnitin will likely show 20% or more.
The colours
Blue — reads as AI.
Purple — borderline. We humanize it too, to be safe.
Papers vary: on about 1 in 5, blue alone catches less than half of what Turnitin flags. Purple closes most of that gap. Measured on 1,707 papers Turnitin scored 20% or more.
The number
Our AI score is typically within 6 points of Turnitin's. That's why the report leads with the verdict.
Good to know
- Turnitin can score the same text differently when it is submitted again.
- Turnitin updates its detector. We re-measure at least every three months.
- Highlights are our estimate. Turnitin may flag other sentences.
How we measure
We run papers that have a real Turnitin result through the same AI Check you use, and compare our verdict, number and highlights with Turnitin's. Copies of a paper count once. We don't publish the papers or their reports.
Our AI Check is an estimate, not Turnitin's result. EliteWorks is not affiliated with Turnitin. Our guarantee