Here is the direct answer. This site is a free AI checker you can run right now. Paste at least 65 words, press check, and you get an AI score from 0 to 100 with one of three plain verdicts: Reads as AI-generated, Borderline, or Reads as human-written. A run costs one credit at any length, and the free tier is 10 credits a day counted per network. There is no account, no login, no card and no email address, so there is nothing to verify and nothing to cancel. GPTZero is a separate AI detection product from a separate company, and this site is not connected to it, affiliated with it, or endorsed by it. Readings from different detectors are not interchangeable, so what you get here is a score from our own checker on our own scale. The rest of this page is what that reading includes and what to do with it.
If you are searching for this, you most likely have a passage in front of you and one question about it: is this going to cause a problem. The checker sits at the top of this page, so you can paste the passage and run it before reading another word.
What people want from a checker tends to be the same short list. No signup standing between them and the answer. No card field on a task that lasts exactly one deadline. No email address to hand over. And some idea of what makes writing read as machine-made, so the reading turns into an edit instead of a worry. That is what this page is built around. The rest of it covers what the score measures, what it does not, and what to do once you have it.
The checker needs at least 65 words. Short passages do not carry enough signal for a stable reading, which is why the floor exists. Above it, length does not change the price. One check is one credit whether you paste 80 words or a full chapter, and that is deliberate. On a long document the sensible first move is to check the whole thing rather than sampling one paragraph and hoping it represents the rest.
The output is a 0 to 100 score plus one of the three verdicts: Reads as AI-generated, Borderline, or Reads as human-written. The words are there so you do not have to guess where a cutoff sits. Borderline means borderline, and that is useful on its own, because a passage in that band usually has a few repeated habits pulling it up rather than a draft that needs replacing.
Because a run costs one credit at any length, you can check the full draft first and then re-check individual sections to find which part is carrying the score. Then edit, then check again. The second check is the part worth insisting on. Editing without re-scoring is how people spend an hour on a passage and never find out whether the hour helped.
This applies to every tool in the category, including this one. An AI checker is a classifier trained on examples. It reports a probability, not a fact about how a document was produced, and it cannot see who typed the sentences. Different checkers are trained on different material, so they can disagree about the same paragraph, and any of them can be retrained over time.
The clearest published measurement of how wrong that can go comes from Liang and colleagues at Stanford HAI in 2023. Across seven detectors, 61% of TOEFL essays written by non-native English speakers were falsely flagged as AI-generated. Those essays were written by people. The finding is not an argument for ignoring scores. It is an argument for reading any score, from any product, as information about a draft rather than a ruling about a person.
That is why the useful next step is never the number by itself. If you want the mechanism behind those false positives, and the page for second-language writers who keep getting flagged.
The registry of AI writing tells lists the specific constructions that make prose read as generated and explains why each one counts. It runs in your browser and costs no credits, so you can work through a draft with it as long as you like. That turns a score into something you can act on, because you are editing a habit you can see rather than nudging a number and hoping.
This is the part that stretches ten credits furthest. Most of what makes a draft read as machine-made is a small set of repeated moves, and you can delete them yourself once you know their names.
The humanizer on this site rewrites every sentence in a passage in one click. One run takes up to 12,000 characters, which is roughly 2,000 words, and needs at least 25 words to work with. It costs one credit per 50 words, so the price follows the size of the job rather than the number of times you press the button.
With 10 credits a day, the arithmetic is worth doing before you start. One check, 400 words of rewriting, and a second check comes to exactly ten. On a longer document, spend one credit checking the whole thing, read the tell registry against the sections that read worst, hand-edit what you recognise, and save the rewrite credits for the paragraphs you cannot see a fix for. Worked examples for specific lengths are at humanize 1,000 words and humanize 5,000 words.
Several other tools here cost no credits at all and have no daily limit, because they run entirely in your browser.
Plenty of people running these checks are not trying to disguise anything. They wrote every word, a course tool disagreed, and now they want to know what set it off. If that is you, check the passage, then check its sections separately to see which part carries the score, and read you wrote it and were accused for what to say and what evidence actually helps.
For the next piece, the drafting record keeps a signed record of the document as it was written, using an ECDSA P-256 signature. Whoever you hand it to can verify without contacting us and without taking our word for anything. Instructors weighing what a detector score is worth as evidence have a page written for that question at for teachers.
One line on policy, then back to the writing. Rewriting a draft you were permitted to generate is ordinary editing. Where AI is not permitted in submitted work, that rule covers the draft however it is edited. Read the policy that applies to your course or your employer, because that is what decides it.
No. There is no account, no login, no card and no email address, and nothing to verify. The free tier is 10 credits a day counted per network, and it refills each day. Because nothing is registered, there is also no subscription to cancel.
A check costs one credit whatever the length, from a 65-word paragraph to a long chapter. A rewrite costs one credit per 50 words. So a day of ten credits is, for example, one check of the full document, 400 words of rewriting, and a second check to see where the score landed.
At least 65 words. Length does not change the price, so checking a whole draft costs the same single credit as checking one paragraph. The humanizer is the one with a size limit: 12,000 characters per run, roughly 2,000 words, with a 25-word minimum.
No, and we make no claim about the output of any other product. Those classifiers are private, they can be retrained by their owners, and different detectors can disagree about the same paragraph. What you get here is a 0 to 100 score and a plain verdict from our checker on our scale.
Reads as AI-generated, Borderline, and Reads as human-written sit alongside the 0 to 100 score so you do not have to invent your own threshold. Borderline is the most useful of the three in practice, because it usually means a few specific habits are pulling the passage up rather than the whole draft being a problem.
Check the passage, then check its sections separately to see which part is carrying the score. Read the registry at AI writing tells for the constructions that commonly cause it, since that one is free and unlimited. The Stanford HAI study by Liang and colleagues in 2023 found that 61% of TOEFL essays by non-native speakers were falsely flagged as AI across seven detectors, so a flag on human writing is a known failure mode. The page at you wrote it and were accused covers what to do next.
Your course policy decides that and nothing else does. Rewriting a draft you were permitted to generate is ordinary editing. Where AI is not permitted in submitted work, that rule covers the draft however it is edited. Read the policy that applies to you.