Claude Detector — Know How Your Text Reads Before Anyone Else Does
Paste any draft and get an instant read on how AI-generated it sounds. Free, 27 languages, no account for your first check — and if it needs work, one click fixes it.
Used by students, writers and editors in 27 languages
Claude Has Started Marking Its Output
Models launched from August 2026 mark what they produce, wherever Claude is offered. Knowing how your own writing reads before you hand it in costs thirty seconds and settles the question.
It travels with your text
Copy and paste does not shake it off. Wherever your text goes, the signal goes with it — so the state your draft is in when you send it is the state that counts.
Your reader is checking too
Professors, editors and clients now run drafts through detectors as routine. Checking first means you see what they will see, while there is still time to do something about it.
Even proofreading counts
You wrote it, Claude tidied it, and the signal is there all the same. That is the case that catches most people out, and the reason to check your own work rather than assume.
The rules changed in August 2026
Models launched from August 2026 mark their output at launch, and Anthropic says older ones are still being brought in. Anything you generate from here on goes out under new conditions.
What You Can Check Right Now
Thirty seconds, and you know where you stand.
- ✓See how strongly your writing reads as machine-written — the same impression a professor, an editor or a client forms in the first paragraph.
- ✓Works on text from any model and on your own writing, across all 27 languages the site supports.
- ✓Get a clear number instead of a vague worry, before you submit rather than after someone else raises it.
- ✓Fix it in one click if the score is not where you want it, without leaving the page.
One thing worth being straight about: this reads your writing, not Anthropic's internal mark — nobody outside Anthropic can read that one yet. Writing is what your reader responds to, and writing is what you can fix in the next minute.
How to Check Your Text
Paste your text
Up to 3,000 words. No account for your first check, no setup, nothing stored or published.
Read your score
You get an AI-likelihood percentage in seconds, so you know exactly where the draft stands.
Fix it if you want to
One click rewrites the text for natural rhythm and specificity, with your meaning and facts untouched.
Frequently Asked Questions
How accurate is it?
It reads writing the way an experienced editor does and gives you a number instead of a hunch — the same signal your reader responds to, made explicit. Treat the score as the verdict on how your draft currently reads: anything above the range you are comfortable handing in is worth one more pass.
Does Claude watermark every response?
Assume yes for anything you generate today. Anthropic says models launched from 2 August 2026 mark their output at launch — across Claude, the API, Claude Code, Claude Cowork and Claude Tag, worldwide rather than only in the EU. Older models are still being brought in, so checking your own drafts is the habit to build now.
Does the watermark survive editing?
Partly — Anthropic says the mark may persist through some editing, and lists heavy editing, paraphrasing, translation and mixing with other writing among the things that leave it undetected. Very short passages carry too little signal to be reliable either way. A full rewrite does what a light edit never will, and you can check the score before and after to see the difference.
Does editing change the result?
Yes — that is the point. Rewrite the flat, evenly-paced sentences and the score moves. Run the check again straight after and you can watch it change.
Should I check my own writing too?
Worth it if Claude touched it at all. Proofreading, translating or summarising your own work through Claude leaves the same signal as text it drafted, so the check tells you something you cannot see by reading.
Why is Anthropic doing this?
Article 50 of the EU AI Act requires providers of generative AI to mark synthetic output in a machine-readable way. Anthropic chose to apply the marking globally rather than only to users in the EU.
Check Your Text in Seconds
Free to start, 27 languages, no account for your first check.
Check My Text — Free
