How the AI detector works
- 1.Paste an assignment or upload a file. One submission can be up to 10,000 characters.
- 2.TutorFlow scores the text and highlights the individual sentences that look generated, so you can see which passages drove the result.
- 3.Read the flagged sentences against what you know about the learner, then decide. The score is evidence for your review, not a verdict.
What the score can and cannot tell you
The detector measures how closely writing matches patterns common in generated text, such as unusually even formality and heavy use of formal connectors. That is a signal, not proof. Careful writing by a strong student can score high, and work by learners writing in a second language is especially prone to false positives. Never treat the percentage as an academic integrity decision on its own.
Questions teachers ask about AI detection
- How accurate is AI detection?
- No detector, including this one, can prove that a person did or did not use AI. A detection score describes statistical similarity to generated text, so both false positives and false negatives happen. Use the result to decide which submissions deserve a closer look.
- Does a high score mean the learner cheated?
- No. A high score means the writing resembles generated text. Talk to the learner about their process, or compare the submission against their earlier work, before drawing any conclusion.
- What does it cost to run?
- The detector uses 1 credit per 1,000 characters, so a 2,000 character essay costs 2 credits.