AI Detector Test Results

This is the log of detector tests for FreeHumanizerAI. Last updated: October 8, 2026. The method and the limits are filled in. The scores are not, because a row only goes in the table after a real sample is tested.

Method

Each finished row should record one sample, the detector that scored it, and the date of that test. The rewrite under test is Ultra mode on FreeHumanizerAI. The original and the rewrite are scored separately.

1. Save the original text and note what kind of sample it is.
2. Count the words in the original.
3. Score the original on one named detector, then score the Ultra rewrite on the same detector.
4. Write down the score before, the score after, and the test date.
5. Add the row only when every cell comes from that test.

Results

TODO(owner): replace the rows below with at least 10 real samples. Include the sample type, the word count, the detector, the score before, the score after, and the test date. Do not publish a score that was not measured.

Limits

Results vary by text and by detector. No humanizer can guarantee 0%. A score on one tool does not transfer to another, and nobody outside Turnitin can test against Turnitin.