Free · no sign-up · no word limits
Make AI drafts read like you wrote them
Check text for signs of AI writing, then rewrite it and re-check every attempt. You see which checks pass, and the editor checks for changes to meaning, quotes and keywords. Review the result before using it.
Rephrases most sentences while aiming to preserve paragraph order and meaning.
This editor uses AI. Checking or humanizing sends text to the server and configured providers; drafts and history may be saved in this browser. Avoid confidential or sensitive information. Review every result; no detector pass is guaranteed.
This acknowledgement is remembered in this browser. Uncheck to stop further submissions.
Free independent check of your input
Check with Scribbr's free AI detector. It uses its own detector and does not provide a Turnitin score.
Copy the text, open Scribbr, then paste and check there. Its free checker accepts up to 1,200 words per submission. Longer text is divided into sections here; check each section separately. These manual results don't return to GPT Cloak or count toward its verification badge.
GPT Cloak sends no text to Scribbr automatically. Pasting there shares it with Scribbr under its privacy terms.
Your rewritten text appears here
Set up a model to humanize, or use the local tidy-up.
Shortcuts: ⌘/Ctrl ↵ humanize · ⇧⌘/Ctrl ↵ check
How it works
- 1
Paste or open your text
No word-count quota. Long documents are processed in sections; very large files may need separate submissions.
- 2
Pick a strength and options
Light, Medium or Heavy. Set the tone, keep meaning strict, and list words that must stay.
- 3
See exactly how it did
Each section is rewritten and checked, and retried a set number of times if a check flags it. You get a clear pass, or a best-effort result that names what still flags it.
What makes it different
Checked before you see it
Most rewriters hand back their first attempt. GPT Cloak checks each rewrite and retries flagged sections a limited number of times. If it still can't pass, it shows you the best attempt and says which checks still flag it.
Every check, shown separately
You see each detector's own answer. Detectors you haven't set up say “Not configured”. The model's estimate is labelled as an estimate. Writing heuristics are tips, never a verdict.
Checks for meaning changes
Strict mode adds safeguards for claims, quotes and numbers. A meaning check compares the rewrite with your original and warns about detected drift. It can miss mistakes, so review the result.
Your keywords survive
List names, product terms or SEO keywords. Each one has to appear verbatim before a rewrite can pass.
See exactly what changed
Compare the result with your original inline or side by side, word by word. Copy it, download .txt or .md, or send it back to the editor.
Free, without subscriptions
No subscription, word quotas or account. Request protections still apply, and AI can pause when the service budget is reached. History stays in your browser, with export available.
Questions
Does this guarantee a detector won't flag my text?
No tool can promise that. Detectors change and disagree with each other. GPT Cloak only says text passed when the applicable configured checks returned a passing verdict. Turnitin is not enabled: it requires licensed integration access and approval. Free manual Scribbr checks use a separate detector and don't count toward the verification badge.
What do the different checks mean?
External detectors are third-party services enabled by the operator; Setup shows their status. The model estimate is the rewriting model judging text on its own, so treat it as an estimate. Writing heuristics are quick local statistics (sentence rhythm, stock phrasing, repetition) that point out things worth fixing.
Is there a word limit?
There is no word-count quota. Longer text is processed in sections and takes longer. Technical request-size limits, timeouts and available capacity apply; very large documents may need separate submissions.
What happens to my text?
Submitted text goes to the server and configured model, plus any external detectors the operator enables. The public service uses Anthropic on Fly.io. The application does not save text in a server history database, but providers have their own retention. Drafts, history and agreement acknowledgement are saved in your browser. Read the Privacy Policy for controls and details.
Which languages work?
Rewrites stay in the language you write in. Quality depends on the model you run: the hosted model handles many languages well, while a small local model is best with English.
Can I stop a long run?
Yes. Cancel asks processing to stop, and your original text stays in the editor. Provider work already underway may still finish; cancellation cannot recall data already sent.
Try it on your own draft
Paste a paragraph and see each check's answer in seconds.
Open the editor