Does AiTextools bypass Pangram? Nobody has asked — and the tool’s entire competitive universe is a detector set that does not include it.
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
- Across a census of all 66 pages on aitextools.com, Pangram appears zero times — including in the site’s own “Top 10 AI Detection Tools 2026” listicle.
- Its contact page says the rewriter is “optimized to produce results that pass each of these”. Its Terms say “We do not guarantee that rewritten text will pass any specific AI detection system.”
- Turnitin is mentioned 1,357 times. The one page dedicated to bypassing Turnitin carries no result claim at all.
- Its six before-and-after claims use six different source texts — climate change, SEO, project management, education, marketing, business — so none can be compared to any other.
- 29 of 32 blog posts are backdated in their machine-readable schema, one to 15 January 2024, almost two years before the domain was registered on 11 January 2026.
- There are no detector screenshots anywhere. The site’s only non-logo image, its social preview, returns a 404.
- It publishes no bypass percentage at all. Its only self-performance number, “95%+ accuracy”, describes its detector, not its humanizer.
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 →Does AiTextools bypass Pangram?
Unknown, and uniquely uncontested. Most pages in this series have to weigh a vendor’s bypass claim against the evidence. Here there is no claim to weigh: AiTextools publishes no bypass percentage at all. Its only self-performance number, “95%+ accuracy”, describes its own AI detector product rather than its humanizer.
What it does say is narrower and, as it turns out, contradicted. We crawled every URL in its sitemap — 66 pages — and counted detector names. Turnitin appears 1,357 times across 66 pages. GPTZero 1,311. Originality 766. Pangram: zero, on zero pages.
This page reports no first-party test. Everything below was fetched on 28 August 2026.
Where do the marketing and the contract disagree?
The contact page states: “The platform checks content against multiple detectors including GPTZero, Originality.ai, Copyleaks, ZeroGPT, and Winston AI. The text rewriter is optimized to produce results that pass each of these.”
Section 16 of the Terms states: “We do not guarantee that rewritten text will pass any specific AI detection system.” Section 14 adds: “Rewritten text may still be detected by some AI detection tools in some cases. We make no guarantees about accuracy, reliability, or effectiveness of our tools.”
It is worth being fair about the size of this gap. AiTextools is not promising 99% and disclaiming everything — its marketing language is “optimized to”, which is softer than most of its competitors manage. But the five detectors it names as optimisation targets are the five it also sells against, and Pangram is not one of them. The optimisation frame is the whole story here: the product was tuned against a detector set that predates Pangram’s relevance.
/pricing.What counts as proof on this site?
Six before-and-after pairs, all rendered as text, none accompanied by an image. And each one uses a different passage.
| Page | Claim | Source text topic |
|---|---|---|
| ZeroGPT humanizer | 100% → 0% AI | The future of AI in business |
| Originality humanizer | 99% → 2% AI | SEO |
| GPTZero humanizer | 98% → 2% AI | Climate change |
| Copyleaks humanizer | 97% → 3% AI | Project management |
| QuillBot humanizer | 96% → 3% AI | AI in education |
| Winston humanizer | 94% → 4% AI | Content marketing |
| Bypass Turnitin | no result claim at all | — |
Six different topics means six different difficulty levels. A detector’s score on a climate-change essay and its score on a project-management explainer are not comparable measurements, so the descending sequence from 100 to 94 tells you nothing about anything.
And note the omission at the bottom. Turnitin is the most-mentioned detector on the entire site — 1,357 times — and its dedicated bypass page is the one with no result to show.
Why are the dates two years out?
The domain aitextools.com was registered on 11 January 2026. The site is seven and a half months old.
Its blog posts say otherwise, but only where a machine reads. 29 of 32 posts carry a JSON-LD datePublished predating the domain’s existence. One is stamped 15 January 2024 — almost two years before registration. Twenty-seven more sit between January and April 2025. The sitemap entry for /terms is dated ten days before the domain existed.
The visible bylines on those pages show plausible 2026 dates. The schema that search engines actually consume does not. The pages affected include the ones that matter most for credibility: the “best AI detectors 2026” ranking, the “best AI humanizers 2026” ranking that places AiTextools first, and a claimed Turnitin accuracy study.
datePublished from all 32 blog posts and compared them against the domain’s RDAP registration record. We requested every image asset directly and recorded its HTTP status. We quote the homepage, contact, pricing and terms pages verbatim, and searched Reddit posts and comments including account-level cross-posting patterns. We also tested whether the site shares templates, numbers or infrastructure with two sibling tools in this cluster; it does not, and we say so below. Where a source has a commercial stake, we say so.Has Pangram named it, and is anyone talking about it?
No, on both counts. AiTextools is absent from the DAMAGE paper‘s 19 named tools and the August 2025 benchmark‘s 20 — both of which predate the domain. The Pangram 4 technical report of July 2026 evaluates 13 commercial humanizers anonymised as “Commercial A” to “Commercial M”, selected by top web traffic in early 2026; a domain registered that January was unlikely to make a traffic-ranked cohort, but that is inference, not a finding.
The community picture is thin and visibly manufactured. Twenty-five Reddit posts exist, several net-downvoted. Seven of them come from one account, which posted the identical title — “Does AItextools actually pass Turnitin?” — to three different subreddits on the same day. The vendor subreddit has 20 subscribers and one post. In the flagship seeded thread, scoring zero, every reply is hostile and nobody confirms the tool works against any detector.
What should you take from this?
AiTextools is optimised against a detector set assembled before Pangram mattered, has never named Pangram, publishes no bypass figure to test, and disclaims passing any specific detector in its own terms. The question in the title has genuinely never been asked by anyone.
What the record supports is Pangram’s own published ceiling: 97.69% of humanized text caught across 13 unnamed tools, with the best evader still caught 92.78% of the time. A tool tuned against GPTZero and ZeroGPT has no particular reason to sit outside that band.
If the real worry is being wrongly flagged on your own writing, keep your drafting history — it costs nothing and beats any rewriting tool as a defence, and 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 AiTextools review. For the Pangram cells with real evidence, see Undetectable AI, StealthGPT, QuillBot, StealthWriter, Phrasly, Humbot, Rephrasy, Ryne AI, Clarity Bubble, SuperHumanizer, UnAIMyText, Text Polish, AuraWrite and HueWrite, plus the Pangram review.
The one tool that got past both
Pangram and Turnitin catch most humanizers — one or the other, and usually the one you did not think to test. StealthWriter is the only tool that bypassed both in 2026. Run Ghost 5.2 Pro at level 7–8, section by section, and re-check before you submit.
Free plan to start, no card. Or get the free prompt pack first →
Affiliate link — I earn a commission if you upgrade.
Frequently asked questions
Does AiTextools bypass Pangram?
Nobody has ever asked, including AiTextools. We crawled all 66 pages in its sitemap and Pangram appears zero times. It is absent from every Pangram publication, and no third party has tested the pair. Its own Terms state that it does not guarantee rewritten text will pass any specific AI detection system.
What bypass rate does it claim?
None. Unusually for this market, AiTextools publishes no bypass percentage anywhere. Its only self-performance figure is “95%+ accuracy”, and that describes its own AI detector product, not the humanizer. What it does claim, on the contact page, is that the rewriter is “optimized to produce results that pass” five named detectors.
Where is the contradiction?
Between the contact page and section 16 of the Terms. The contact page: “The platform checks content against multiple detectors including GPTZero, Originality.ai, Copyleaks, ZeroGPT, and Winston AI. The text rewriter is optimized to produce results that pass each of these.” The Terms: “We do not guarantee that rewritten text will pass any specific AI detection system.” Both live on the same site, both current.
What is wrong with its before-and-after numbers?
They are text-only and uncontrolled. Six detector pages each publish one pair — ZeroGPT 100 to 0, Originality 99 to 2, GPTZero 98 to 2, Copyleaks 97 to 3, QuillBot 96 to 3, Winston 94 to 4 — and every one uses a different source text on a different topic. There is no shared control passage, no date, no generating model, no repeat trials, and no screenshot. And the bypass-Turnitin page, for the site’s most-mentioned detector, carries no result pair at all.
What is the date problem?
The domain was registered on 11 January 2026. Yet 29 of its 32 blog posts carry JSON-LD datePublished values predating that — the earliest 15 January 2024, nearly two years early — including the pages that rank detectors and claim a Turnitin accuracy study. The visible bylines show plausible 2026 dates; the machine-readable schema that search engines consume does not.
Is AiTextools in Pangram’s published research?
No, in all three sources. The July 2026 Pangram 4 report tests 13 humanizers anonymised as “Commercial A” to “Commercial M”, selected by top web traffic in early 2026, so presence cannot be strictly ruled out — though a domain registered that January was unlikely to qualify for a traffic-ranked cohort. Absent from the named lists is the accurate phrasing.
What does it cost and can I get a refund?
Free, Lite at $9.99, Pro at $12.99 and Exclusive at $19.99 a month, with humanizer input capped between 30 and 300 words per request. The refund terms contradict themselves on the same page: the clause is headed “24-Hour Refund Window” while the summary above it describes a “3-day refund policy” in the same sentence as “within 24 hours.”
Who operates it, and is anyone using it?
No legal entity, jurisdiction, registration number, VAT number, address or named person appears on any of the 66 pages — only a support email, behind a Cloudflare-redacted registration. Reddit shows 25 posts, seven of them from a single account that posted the identical title to three subreddits on one day, plus a 20-subscriber vendor subreddit with one post. Nobody in those threads confirms the tool works.
