Check if text was written by AI.
Rewrite it so it sounds human.
Two free tools. The detector estimates how machine-like a piece of text looks. The humanizer rewrites it with varied rhythm, contractions, and natural phrasing — without adding facts.
Free · Sign in with Google · 10 checks + 10 humanizes a day
Sign in to use the tools
The detector and humanizer need a free account so we can give you your daily limit. Sign in with Google — it takes one tap.
What we collect, plainly: when you sign in with Google we store your email address, your name, and how many detector checks and humanizes you use each day. That is all. We use it to run your account and enforce the free daily limit. We do not track you anywhere else, and we do not sell your data. Your pasted text is processed in your browser and is never sent to our servers.
🔍 AI Detector
Paste text and get an AI-likelihood score, a per-sentence breakdown, and the signals behind it — perplexity and burstiness, measured with a small language model running in your browser.
Open the detector →Sign-in required · 10 free checks a day
✍️ AI Humanizer
Paste AI-sounding text and get a rewritten version in a casual, professional, or academic tone. Then check the result with the detector in one tap.
Open the humanizer →Sign-in required · 10 free humanizes a day
How it works
- Detector: a GPT-2 model runs in your browser and measures how predictable each sentence is (perplexity), plus how much sentence length and predictability vary (burstiness). Very predictable, very uniform text scores higher. The first check downloads the model once; after that it runs locally.
- Humanizer: a rule-based rewrite — contractions, varied synonyms, split and merged sentences, varied openers — in the tone you pick. No AI API is involved, so nothing you paste leaves your browser.
- Limits: 10 detector checks and 10 humanizes per day, per account, counted on our side. No payment, no card, no upgrade screen.
Read this before you trust either tool: the detector is a heuristic estimate, not a verdict — no detector can prove who wrote a text, and all of them get it wrong sometimes. The humanizer changes style, not certainty — nothing here guarantees passing any detector, and no tool can make text “undetectable.” Use both as aids, not as proof.