Can noredink detect AI?
No. NoRedInk has no AI detector and produces no AI percentage. Its Originality Insights feature reports process data instead — time spent per draft, copy-paste events, and pasted-text percentage — and NoRedInk states explicitly that it does not judge whether a writing sample is authentic. Teachers interpret; the tool doesn't accuse.
Why — the first-principles explanation
NoRedInk made a design choice that is quietly smarter than what most of the industry does, and understanding it explains the whole product.
Everyone else tries to answer "does this text look like AI?" That question has no good answer, because text carries no origin signal. Words are just words; nothing is embedded. So detectors fall back on scoring perplexity — how predictable the writing is — and the proxy misfires badly. Stanford researchers found seven detectors flagged 61.22% of TOEFL essays by non-native English speakers as AI-generated while classifying U.S. eighth graders near-perfectly. The tool measures simplicity and reports it as cheating.
NoRedInk asks a different question: "how did this text get into the box?" Because students write inside NoRedInk's editor, the platform can observe the process directly — when text appeared, how long the student spent, whether prose was typed or arrived in a paste. That's not inference. That's a log. And it sidesteps the bias entirely: a slow, careful ESL student's typing history looks exactly like a slow, careful native speaker's typing history, because keystrokes have no accent.
The consequence is the important part, and NoRedInk says it plainly: Originality Insights does not judge authenticity. It reports that 400 words arrived in a single paste at 11:58 p.m. It does not say those words came from ChatGPT. They might have come from the student's own phone notes, a shared Google Doc, or an outline they wrote last week. NoRedInk's own guidance is that pasted text isn't automatically plagiarism, and that teachers are the ones positioned to judge.
That's a feature, not a limitation. A detector hands a teacher a number that feels like a verdict and isn't one — which is how false accusations happen at scale. Originality Insights hands the teacher a fact and leaves the judgment where it belongs. NoRedInk's separate Grading Assistant uses AI to help score and give feedback on writing; it doesn't evaluate whether the work is original either.
An example that makes it click
Two ways for a teacher to check whether you did your own math homework. One: stare at your answers and decide if they look "too good." That's an AI detector — and it means the kid with naturally neat work gets accused every time.
Two: notice that the whole worksheet was blank at 11:57 and completely filled at 11:58. That's Originality Insights. It doesn't say you cheated. Maybe you copied it over from scratch paper you'd been working on all evening. It just says: this appeared all at once. Then your teacher — who knows you, and can ask — figures out which story is true.
How to do it
- Teachers: open the student's submission in NoRedInk and look at Originality Insights, which reports time spent per draft, copy-paste detection, and the percentage of text that was pasted.
- Read a high paste percentage as a question, not an answer. NoRedInk states the insights do not detect plagiarism and do not judge authenticity — pasted text may be the student's own notes or an earlier draft.
- Combine paste data with draft time. Text that appeared instantly with almost no editing time is more informative than either signal alone.
- Ask the student before concluding anything. NoRedInk's guidance is that teachers are best positioned to determine whether pasted text is a problem.
- Students: compose inside the NoRedInk editor rather than writing elsewhere and pasting in, even when the words are entirely your own. A single large paste looks the same as any other large paste.
- Students: if you draft on paper or in another app first, tell your teacher up front rather than after the report flags it.
Key facts
- NoRedInk's Originality Insights reports time spent per draft, copy-paste detection, and pasted-text percentage — not an AI probability score.
- NoRedInk states that Originality Insights does not judge student writing samples for authenticity and that the insights do not detect plagiarism; teachers are positioned to decide whether pasted text is a problem.
- NoRedInk's Grading Assistant is a separate AI-powered feature focused on scoring and feedback; it does not evaluate whether student work is original.
- Seven commercial AI text detectors flagged 61.22% of TOEFL essays by non-native English speakers as AI-generated while classifying U.S.-born eighth graders' essays near-perfectly (Liang et al., Patterns, 2023) — the bias that process data like paste tracking avoids.
- The FTC alleged Workado marketed its AI Content Detector as 98% accurate when independent testing showed roughly 53% on general-purpose content; final order approved August 2025.
▶ The 60-second explainer (script)
Can NoRedInk detect AI? No — and the reason is more interesting than the answer. NoRedInk has no AI detector. There's no AI percentage anywhere in it. What it has is Originality Insights, and that feature asks a completely different question than every other tool on the market. Everyone else asks: does this text look like AI? That question has no good answer, because words carry no trace of who wrote them. So detectors guess at style — they flag writing that's too predictable. Stanford tested seven of them: sixty-one percent of essays by non-native English speakers came back flagged as AI. The tool was measuring simple vocabulary and calling it cheating. NoRedInk asks instead: how did this text get into the box? Since students write inside its editor, it can just watch. Time spent per draft. Copy-paste events. What percentage of the words were pasted. That's not inference — it's a log. And it dodges the bias completely, because keystrokes have no accent. A careful ESL student's typing history looks exactly like anyone else's. Here's the crucial part, and NoRedInk says it right in their docs: Originality Insights does not judge authenticity. It reports that four hundred words landed in one paste at 11:58 PM. It does not say those words came from ChatGPT. Maybe they came from the kid's phone notes. Maybe from an outline they wrote last Tuesday. The tool gives the teacher a fact and leaves the judgment to a human who can just ask. That's not a weaker product. That's the honest one.
What authoritative sources say
People also ask
Does NoRedInk know if I used ChatGPT?
It cannot know. It can see that text was pasted rather than typed, and how long you spent in the draft. Where pasted text came from is invisible to it — that's why NoRedInk says the insights don't judge authenticity.
Will NoRedInk flag me if I paste my own writing?
Originality Insights will report the paste, because it reports all pastes. It won't call it cheating. NoRedInk's guidance is explicit that pasted text isn't automatically plagiarism and that the teacher decides.
Is NoRedInk's Grading Assistant an AI detector?
No. The Grading Assistant is an AI tool that helps teachers score writing and generate feedback. It evaluates the quality of the work, not whether the student actually produced it.
Why doesn't NoRedInk just add an AI detector?
Style-based detection carries a documented bias — seven detectors flagged 61.22% of non-native English essays as AI in Stanford testing — and the FTC has taken action over inflated accuracy claims. Process data avoids both problems by reporting what happened instead of guessing.