What "Verified human" means

"Verified human" is the mark a check gives a text when the standard detector reads it as human. Verified human: the standard detector read this text as human.

It covers this text on this date. It does not show who wrote it or how.

The mark is for words a person wrote, as handed in. Text rewritten by a humanizer does not qualify, even if a person wrote the first draft. Our detector does not catch every rewrite: in our tests it missed most rewrites by one tool.

Verified human: the standard detector read the text as human.

Not verified: part of the text read as machine-written, or a check that decides the mark did not pass. The result shows the sentences and the reason.

Not scored: Not enough text to read. A check needs at least 50 words.

Couldn't finish: a reading did not complete. Run it again in a minute.

Our standard detector flagged 0 of 600 human-written papers in its release test (95% upper bound 0.50%), 1 of 549 held-out human papers, as Mixed (upper bound 0.86%), 5 of 9,305 real student essays, all as Mixed, none as AI, and missed 0 of 481 wholly machine-written held-out papers (upper bound 0.62%).

It called 20 of 176 humanized AI papers human and, in a newer test of one humanizer tool, 231 of 255 rewritten long papers and 122 of 136 rewritten essays human, so a humanizer tool can get text past it.

A second detector, trained on student essays, also reads every text the standard detector calls human. Its reading is shown as a note beside the result and does not decide the mark.

It never says a text is AI-written as a fact about a person, and it never gives a probability that anyone cheated.

A flag should be reason to look closer and not verified proof of cheating.

Every human-written document in its tests was written before 2022.