Copyleaks  Reviews 

J’ai confié à une boîte à soupçon — un détecteur d’IA de pacotille — des textes écrits entre 2005 et 2025. Rappel utile : en 2005-2006, le grand public n’avait pas d’IA sous la main. Pourtant, je lance la machine sur un texte de 2006 (le mien, pas celui d’un clone), et paf : 90 % IA. On rit. Je relance : 84 %. Une troisième fois : encore moins. Plus je remue, plus l’oracle en plastique change d’avis. Solide, n’est-ce pas ?

Je passe d’autres textes jamais écrits sous IA : verdicts en rafale 80 à 90 % IA. Conclusion délicieuse : je serais donc l’égal de l’IA ; à quoi bon en utiliser une ? Et inversement, puisque l’engin me colle 90 % de sa patte, pourquoi m’éreinter à écrire, si n’importe quel quidam peut m’accuser d’être un faux auteur au nom de son pourcentage magique ? Dilemme ? Non : scandale, et potentiellement diffamatoire.

Cerise lucrative : l’officine vend l’abonnement au soupçon et l’exorcisme payant. On vous tamponne « IA », puis on vous revend, contre une autre redevance, un texte « humanisé ». Par quoi ? Par… de l’IA. Le gendarme confisque la came, ouvre sa boutique et vous la revend en souriant. Monde à l’envers, mais caisse enregistreuse à l’endroit.

Si ce cirque avait été gratuit, j’aurais pris leur prose « humanisée », je l’aurais resoumise comme texte original… Devinez la réponse. On touche ici à un risque très concret : chasses aux sorcières en entreprise, association, confrérie. Il suffira d’un zélé porteur d’écran pour brandir 85 % et pourrir une réputation. On ne lit plus : on affiche des chiffres.

J’invite donc à tenir ce mécano à distance respectable. Fumisterie rentable, dommage civique.

NB. Pour aller au bout : j’ai soumis ce commentaire au bidule. Verdict : 92 % IA, 8 % humain. Puis j’ai demandé à ChatGPT 5 un condensé « balzacien ». Re-test : 97 % IA. Question naïve : pourquoi cette société a pignon sur rue ? Et pourquoi les avis 5 étoiles restent invisibles ?

Conclusion : rigolade, mais scandale. Des logiciels du genre, il y en a une grappe ; ce qui ne varie pas, c’est l’accusation à l’emporte-pièce et ses pourcentages catéchétiques.

On vit une drôle d’époque, non ?

The tool was decent at detecting AI content, but you could easily fool it by changing a few words around. Service will not make corrections if you make a mistake choosing plans or upgrading/downgrading, so do so carefully.

No da un porcentaje real, solo dice 0% o 100%. Es inútil para detectar qué partes del texto han sido generados por IA, como sí hacen otras plataformas. Puedes escribir tú mismo frases entre medias de un texto generado por IA y te dirá que lo tuyo también es IA. O puedes intercalar partes generadas por IA en un texto amplio y te dirá que hay 0% presencia de IA.

This is still crappy. I try to give it the benefit of the doubt, but nope. I have 35 years of writing experience, and I write things on a whim in a journal. I put it in CopyLeaks, and it says AI Content 100%. I put it into any other software, and it says 100% Human. CopyLeaks...Epic fail!

Expensive, not truly effective.

And most importantly, and reason why it's 1 star rather than 3, is how difficult they made to deactivate the account.

I signed up to CopyLeaks for what I assumed would be a straightforward plagiarism detection service. I paid $16 and was given 100 credits, expecting that to be sufficient for my needs. However, the service is extremely misleading in both functionality and billing.

The first scan I performed was completely inaccurate. It flagged everything in my document as plagiarised simply because it wraps all content in quotation marks, regardless of whether it’s actual quoted material or just standard speech formatting. This inflated the plagiarism score and forced me to run another scan.

What I wasn’t clearly told is that every new scan consumes additional credits. Even worse, the second scan flagged my document as 100% plagiarised simply because it matched the first version I had submitted — which the system had stored in its own database. So essentially, I was penalised for checking my own work twice.

There is no clear guidance or explanation on how to properly use the service or configure settings. By the time I had figured out how to do a clean, correct scan, I had used up all my credits. To proceed, I was required to top up my account by the minimum purchase amount — another $16 — meaning I had to pay twice for a single functional scan.

This is, in my opinion, an unfair business model. It’s not transparent, it penalises genuine users trying to check their work, and it locks essential settings behind trial and error at a real cost.

I strongly advise others to look for a more honest and user-friendly plagiarism detection service.

Please don’t believe in this tool! Flags everything as 100% AI .. even my books I’ve never published but I wrote them in 2014 also shows AI but clearly is not AI.
But if you buy a paid version than its 0% or only 10% FAKEEEE

FAKE. DON'T RELY ON IT ALONE

Un texto escrito por mi me lo marca 100% escrito por IA. Y no me dice que partes y por qué, pues tengo que comprar la suscripción... blanco y en botella.

As an experienced writer navigating the industry for over a decade, I've unfortunately encountered the most frustrating plagiarism and AI detection tool in my career. Each time I revise my work, this tool inexplicably flags different content, which is not only baffling but also incredibly disruptive to my workflow.

Last Thursday, I raised a ticket to address these persistent issues, only to receive an automated response. Nearly a week has passed, and still, there’s been no resolution from their support team, leaving me feeling neglected and unheard.

To complicate matters, my clients are insisting that I rely on this flawed tool without question. Despite providing them with three separate reports from Copyleaks indicating the consistency of the content, they remain unconvinced and deaf to my concerns. It's utterly exhausting, draining my time, energy, and finances. I find myself checking each piece of content multiple times - often 8 to 9 iterations - just to appease my clients.

I can’t stress this enough: please think carefully before investing your hard-earned money in subpar tools like this one. It’s been nothing short of a headache.

Would give 0 stars if I could. It is so hypersensitive that it has flagged writing of mine that was written before the existence of generative ai. Fake ai detector, needs to go bankrupt.

Probably one of if not the worst ai detectors out there. No matter what you put it it marks it as AI and ive tested it multiple times.

Completely useless, fake AI detector, flags everything as AI even when it's clearly not.

Long explanation:
I've tested it by inserting first 3 AI generated texts, flagged them as AI, then I thought "ok this might work".
Inserted 5 texts that I wrote myself, flagged them as AI, which is dumb because none of it was AI in any way.
Then I've decided to copy paste very bad slurs that AI will refuse to say under any circumstance, a single slur repeated over 300 times, flagged it as AI (when AI can't even say that stuff).

This past week, this detector has been flagging everything as AI, I even check personal emails I sent my family. How can this service be even allowed to get used by businesses?! This is ridicilous.

The most inaccurate AI detector!! I encourage everyone to use zerogpt. This one is completely useless. And the 5-star ratings it has sound generic AF, so I'm 100% sure they are not from legit users.

I think this Ai detector is just promoting itself as the most accurate one. But the accuracy is zero. Out of 10 human written articles, 9 are flagged as AI. How can this detector be so reliable. I just don't understand what are your parameters and how do you check. I think they are just fooling people. Even a simple line is detected as AI. Rubbish. Companies should not rely on Copyleaks.

Abysmal. The software just doesn’t work. A simple sentence structure used in everyday sentence will get flagged. Furthermore they have an egregious refund policy, if you use a single credit (which you’ll have if you don’t do your due diligence and have to find out first hand it doesn’t work) they don’t offer refunds irrespective of if you want to cancel 2 minutes after subscription. Stay away do not use. (If you run the positive reviews on this site through copyleaks it’ll declare it Ai. Take what you will from that)

very bad always trying to verify if you are a robot, bad login ui keeps looping, always trying to get you to pay for a simple scan.

This is a big scams
We write himself, copyleaks show AI content, SHAME SHAME SHAME SHAME