rel=nofollow citations. The promoted call-to-action above is a StealthWriter affiliate placement carrying its own disclosure, and is unrelated to this review.Does Rephrasy bypass GPTZero? It guarantees 100% in four places, and one of those four places is the disclaimer that takes it back.
The only tool that bypassed Pangram and Turnitin in 2026.
Most humanizers clear one detector and get caught by the other. StealthWriter is the one that gets past both — run Ghost 5.2 Pro at level 7–8, section by section, and re-check before you submit.
Try StealthWriter →Key Takeaways
- Rephrasy promises the same thing in four places and withdraws it in the fourth. Its homepage says “We guarantee 100% pass rates on all major AI detectors.”
- That guarantee is also submitted to Google as machine-readable FAQ markup, naming GPTZero by name: “Rephrasy bypasses all major AI detectors including Turnitin, GPTZero… We guarantee 100% pass rates on every major detector.”
- Its Terms then contradict themselves six sections apart. Section 1 says the service makes text “undetectable by AI detectors.” Section 7 says it “does not guarantee that humanized text will be undetectable by AI detectors.”
- The Terms carry no last-updated date at all, and contain zero mentions of academic integrity, cheating or students — on a site headlined for students.
- There is no GPTZero screenshot anywhere. The entire image inventory is three use-case graphics and a university logo.
- Its structured data declares 4.9 stars from 105 reviews. Neither number appears anywhere in the visible page text.
- Unusually for this market, it does publish a real legal entity: a German address, a VAT number, and German governing law. Credit where it is due.
Does Rephrasy bypass GPTZero?
No published evidence says so, and the interesting part is where the promise falls apart. It is not the usual gap between a loud homepage and a quiet contract. Here the contract disagrees with itself.
Rephrasy names GPTZero in its marketing, in its structured data and in its Terms. It shows a GPTZero result in none of them.
This page reports no first-party test. Everything below was fetched from rephrasy.ai on 21 September 2026 and quoted verbatim.
Where does the promise appear?
Four places, on one website, on one day.
| Where | What it says | Audience |
|---|---|---|
| Homepage copy | “We guarantee 100% pass rates on all major AI detectors.” | Readers |
| FAQPage structured data | “Rephrasy bypasses all major AI detectors including Turnitin, GPTZero… We guarantee 100% pass rates on every major detector.” | Search engines |
| Terms, section 1 | “humanizes AI-generated text, making it undetectable by AI detectors” | Users, contractually |
| Terms, section 7 | “does not guarantee that humanized text will be undetectable by AI detectors” | Users, contractually |
Why does the markup version matter more?
Because it is written for machines, not people.
The 100% guarantee is not only in body copy. It sits inside the page’s FAQPage JSON-LD, the structured data submitted so search engines can lift questions and answers straight into results. That is the copy Google parses, and it is where GPTZero is named explicitly alongside Turnitin, Originality.ai, Copyleaks, ZeroGPT, Sapling and Content at Scale.
No sample size, corpus, model, test date or screenshot accompanies it in either location.
The same page does something comparable with its reputation. Its structured data declares an aggregate rating of 4.9 from 105 reviews, and neither number appears anywhere in the visible text. We found the identical pattern on GPT Scrambler, which declares 1,250 ratings no reader can see. Two unrelated vendors, the same choice about who the number is for.
Can you check GPTZero yourself?
Turnitin sells no individual licences and shows students no AI report, so nobody outside an institution can produce a genuine Turnitin screenshot. GPTZero removes that excuse entirely. Anyone can paste up to 10,000 characters into gptzero.me right now with no account, no card and no signup; a free account raises the per-scan limit to 150,000 characters with a 10,000-word monthly quota, and paid tiers start at $9.99 a month.
So when a humanizer claims to beat GPTZero and shows you nothing, that is a choice, not a constraint.
It also means you can check your own text. A genuine current GPTZero result shows a three-class classification — human, AI or mixed — a confidence category rather than a bare percentage, a probability breakdown across all three classes, and sentence-by-sentence highlighting. An image showing only “X% AI”, or one displaying perplexity and burstiness, is either fabricated or dates to roughly 2023.
Which is what makes the empty proof wall conspicuous. Rephrasy publishes no detector screenshots at all. Its complete image inventory is three use-case graphics and a university logo.
For Turnitin there is a real excuse, and we have made it on Rephrasy’s behalf elsewhere: Turnitin sells no individual licences and its AI report is instructor-facing, so no vendor can lawfully produce one. GPTZero hands that excuse back. A scan costs nothing, needs no account, and would produce the one artefact that settles the question.
What do the Terms actually commit to?
Very little, and they do not say when they last said it.
There is no last-updated date anywhere on the Terms page. Section 7 disclaims the outcome. Section 5 says refunds are possible under specific conditions but that “refunds for already used credits are not available”, with exceptions case by case — and the headline product is excluded outright by the line “We do not refund any lifetime deals!” on a Lifetime Access tier sold at $199.
The Terms also contain zero mentions of academic integrity, cheating, students or educational institutions. That is notable on a site whose own headline reads “The #1 AI humanizer trusted by 200,000+ students & professionals” and whose pitch is to “bypass Turnitin, GPTZero and all AI detectors” before you submit an essay. Most tools in this cluster carry such a clause even while marketing against it.
What does Rephrasy do better than its rivals?
One thing, clearly, and it should not be buried under the criticism.
It publishes a real legal identity. The Terms name a provider address at Stettiner Str. 41, 35410 Hungen, a VAT number, and state that the terms are governed by German law. If something goes wrong there is a jurisdiction and an entity to point at.
Set that against the rest of this cluster. Clarity Bubble publishes no entity, no jurisdiction and no address, only a Gmail address. Text Polish gives its governing law as “the jurisdiction in which TextPolish operates”, which it never names. HueWrite carries SOC 2 and PCI DSS badges with no auditor, report or date.
Being findable is not the same as being right. It is still a real difference, and it is the reason the contradiction in its Terms is worth taking seriously rather than shrugging at: this is a company that can be held to what it writes.
Has any detector company tested it?
None has. Rephrasy is absent from the appendix table in GPTZero’s February 2026 paper, which names nine bypass services individually; absent from the nineteen tools audited by name in Pangram’s DAMAGE paper; and absent from the twenty scored in Pangram’s August 2025 benchmark — despite Rephrasy running its own dedicated page about bypassing Pangram. We cover that in does Rephrasy bypass Pangram, and the Turnitin side in does Rephrasy bypass Turnitin.
How reliable are the numbers on either side?
GPTZero’s own FAQ concedes: “Our classifier is not trained to identify AI-generated text after it has been heavily modified after generation.” Its developers page simultaneously claims a “Paraphraser Shield” means “even if AI content has been altered to look more human-like, GPTZero can detect it.” Both statements are currently published.
The two leading detector vendors publish opposite numbers on the same question. GPTZero’s February 2026 paper reports 93.5% recall on 1,000 texts run through nine bypasser services, and puts Pangram at 49.7%. Pangram’s DAMAGE paper reports GPTZero at 60.04% on humanized text, and 34.53% at GPTZero’s own default threshold. Each benchmark shows its publisher winning, and neither is peer-reviewed.
On false positives, GPTZero was one of seven detectors in Liang et al. (2023), which reported a 61.22% average false positive rate across those seven on 91 TOEFL essays. No per-detector figure was ever published, so that number is not GPTZero’s. GPTZero now claims it has cut its own TOEFL false positive rate to 1.1% — a figure no third party has verified.
What should you take from this?
Rephrasy is a real company with a real address that guarantees something its own disclaimer refuses to guarantee, in markup written for a search engine, about a detector it has never shown itself beating.
The honest reading of section 7 is the accurate one: nobody, Rephrasy included, can promise your text will come back human. Section 1 should not say otherwise, and the homepage should not put a number on it.
The check costs a minute and nothing else. Paste your own output into gptzero.me, free, no account, up to 10,000 characters, and read the three-class verdict with today’s model version stamped on it. Then keep your drafting history, because that is what protects you whichever way the detector goes.
If your real worry is being wrongly flagged on your own writing, keep your drafting history — it costs nothing and it survives a detector being wrong. It matters most for the writers detectors treat worst: ESL writers and AI detection and false positives for neurodivergent students. Check before you submit with the best pre-submission check and the Turnitin self-check, and strip the obvious tells by hand first via what to remove before using an AI humanizer.
For the product itself see our Rephrasy review, and the same question against Turnitin and Pangram. Also in this cluster: GPT Scrambler, RewriteIQ, GenZWrite.
Get the free prompt pack instead
The manual rewriting prompts that lower AI signals without a subscription — and without betting a submission on a number nobody has reproduced. Free, no card.
Frequently asked questions
Does Rephrasy bypass GPTZero?
Nothing published establishes it, and the company’s own contract declines to promise it. Rephrasy guarantees 100% pass rates on all major AI detectors in its marketing and names GPTZero explicitly in its structured data, but publishes no sample size, no test date, no model, no scan mode and no screenshot. Section 7 of its Terms states that it does not guarantee humanized text will be undetectable. GPTZero has never tested Rephrasy, and no independent party has published a result on it.
What is the contradiction inside the Terms?
Section 1 and section 7 say opposite things about the same product. Section 1, Purpose and Scope, reads: Rephrasy provides a web application and API that humanizes AI-generated text, making it undetectable by AI detectors. Section 7, Disclaimer, reads: Rephrasy aims to humanize text effectively but does not guarantee that humanized text will be undetectable by AI detectors. Both were live on 21 September 2026. Most vendors in this series contradict their marketing in their contract. This one contradicts its contract in its contract.
Why does the structured data matter?
Because it is the version search engines read. The 100% guarantee is not only in body copy; it is inside the page’s FAQPage JSON-LD, the machine-readable markup submitted for search results, where it names GPTZero among the detectors bypassed. A claim placed there is aimed at Google rather than at a reader, and it carries no sample size, corpus, model, test date or screenshot.
Is there any evidence at all?
None specific to GPTZero. The site publishes no detector screenshots of any kind; its complete image inventory is three use-case graphics and a university logo. This matters more for GPTZero than for Turnitin, because GPTZero takes 10,000 characters with no account and no card, so a screenshot costs nothing but a minute. Turnitin sells no individual licences, so no vendor can lawfully produce a real Turnitin report; GPTZero offers no such excuse.
What about its ratings?
They live where search engines look rather than where readers do. The homepage’s structured data declares an aggregate rating of 4.9 from 105 reviews. Neither 4.9 nor 105 appears anywhere in the visible text of the page, and no review surface on the site collects a rating. We found the same pattern on GPT Scrambler, which declares 1,250 ratings that no reader can see.
Does it restrict academic use?
No. The Terms contain zero mentions of academic integrity, cheating, students or educational institutions, while the site is headlined as the #1 AI humanizer trusted by 200,000+ students and professionals. Most competitors in this cluster do carry such a clause even while marketing against it. Rephrasy does not, and its Terms carry no last-updated date to say when that was decided.
Is anything about it better than its rivals?
Yes, and it is worth saying. Rephrasy publishes a real legal identity: a German provider address at Stettiner Str. 41, 35410 Hungen, a VAT number, and terms governed by German law. Across this cluster that is rare. Clarity Bubble, Text Polish and HueWrite publish no entity, no jurisdiction and no address between them. Being findable is not the same as being right, but it is a real difference.
What does it cost, and can I get a refund?
A free plan humanizes up to 100 words. Quick Pass is $12.99 for 24 hours, Growth $99 a year, Business $120, Professional $220, and Lifetime Access $199 reduced from $300. On refunds the Terms say refunds are possible under specific conditions, that refunds for already used credits are not available, and that exceptions are case by case. The headline product is excluded outright: we do not refund any lifetime deals.
