Free AI detector that shows its working
Check if text is AI generated, sentence by sentence. See the rhythm of your writing, the exact words that look machine-made, and the evidence both ways – not just a number.
The scan runs in your browser. Nothing is uploaded unless you ask for the AI second opinion.
Your result appears here: an overall score, the rhythm trace of your sentences and every sentence lit by how many AI-typical features it carries.
What the detector actually measures
Language models write by choosing likely next words. That leaves fingerprints you can count without any AI at all, which is why this check runs instantly in your browser:
- Burstiness. People write in bursts: a four-word sentence, then a forty-word one. Models keep a steady 15–25 words. The rhythm trace draws this as a seismograph line.
- Stock vocabulary. Words such as “delve”, “tapestry”, “pivotal” and “underscore”, and phrases like “it’s important to note”, appear far more in model output than in ordinary writing.
- Formula cadence. Sentences that open with Furthermore / Moreover / Additionally, tidy lists of three, and paragraphs of identical length.
- Missing human texture. Few contractions, no questions or asides, no brackets, no specific numbers or personal detail.
Each sentence gets its own score, so you can see where the signals cluster. A pasted AI paragraph inside a human essay lights up on its own.
Read how AI detectors work, with a game that shows what “perplexity” means.
Try it: rhythm, not vocabulary
Switch between a real-sounding student paragraph and a model-style one. Watch the trace.
- Variation (CV)
- 0.67
- Shortest
- 2
- Longest
- 23
Short bursts (“Small win.”) sit next to long, rambling sentences. The trace jumps around.
ahaboo offers a small number of guided explainers grounded in trusted sources, where you drag and zoom your way through how things work.
A score is a reason to look closer, not a verdict
No detector can prove authorship. OpenAI shut down its own AI classifier in 2023, reporting that it correctly identified only 26% of AI-written text while wrongly flagging 9% of human text. Turnitin states that its AI score should not be used as the sole basis for action. Researchers at Stanford showed that detectors are biased against non-native English writers.
So use the result the way an examiner uses a UV lamp: it shows you where to look. If a passage glows, compare it with the writer’s earlier work, ask for drafts or version history, or talk it through. The accuracy guide has a simulator that shows why even a “98% accurate” detector wrongly accuses people when most of a class is honest.
Who uses VerifyWrite
Teachers triaging a pile of essays use the essay detector and the one-click essay checker. Students check their own work before submitting, so an over-polished paragraph doesn’t cause a misunderstanding. Editors screen freelance copy and run the originality search before publishing.
Questions people ask
Is this AI detector free?
Yes. The detector, the rhythm trace and every counting and readability tool are free with no account and no word limit beyond what your browser can handle. The optional AI second opinion is also free, with a fair-use limit of a few requests every ten minutes.
How do I check if text is AI generated?
Paste at least 40 words (ideally 300 or more), press “Check for AI signals” and read three things together: the overall score, the rhythm trace of sentence lengths and the highlighted sentences. Then ask for the AI second opinion if you want the evidence spelled out. Treat all of it as a reason to look closer, never as proof.
Can an AI detector prove that someone used ChatGPT?
No. Every detector, including commercial ones, estimates likelihood from statistical patterns. OpenAI withdrew its own AI text classifier in July 2023 because of its low accuracy. A fair decision needs other evidence: drafts, version history, notes, or a conversation with the writer about the work.
Does VerifyWrite store what I paste?
The detector runs entirely in your browser, so the text never leaves your device. If you click “Get a second opinion”, that one passage is sent to our AI provider to be analysed and is not stored by us. Your recent scans are saved only in your own browser’s local storage, and you can clear them any time on the History page.
Why does my own writing score as AI?
Formal, carefully edited prose with even sentence lengths, no contractions and textbook transitions looks statistically similar to model output. Non-native English writers are affected most: a 2023 Stanford study (Liang et al.) found popular detectors misclassified over half of TOEFL essays by non-native speakers as AI-generated.
What languages does it support?
The signals are tuned for English. Other languages will produce a score, but vocabulary and contraction signals won’t mean much, so treat non-English results as unreliable.
The rest of the kit
Your text follows you between tools, so you can paste once and check everything.