Turnitin's AI score is an estimate, not proof. It hides scores between 1% and 19% because low scores are less reliable, a 2026 study found it gave 0% to papers written fully by the newest AI models, and human writing can still be flagged.
Written by Adithyan, a student who builds PaperCheck. Last updated 31 October 2026.
In a 2026 study of 160 academic papers (Van Vlasselaer and others, International Journal for Educational Integrity), Turnitin scored papers written fully by the newest models at 0% on average, while other detectors caught most of them. Detectors disagree because they are trained on different AI writing.
Formal, polished or non-native English writing can read as AI to some detectors. That is why a score alone should never decide a case.
Yes. It can miss text from new models and can flag human writing. Its score is an estimate.
The AI score is between 1% and 19%, which Turnitin does not show exactly.
Check your paper: 5 free checks a month
PaperCheck is not affiliated with Turnitin, GPTZero, papercheck.in or papercheck.ai. Scores from any detector are estimates, not proof.