Paste any text for a human-read score, the exact AI phrases it contains, per-sentence highlighting, readability and stylometry, and a free one-click clean-up. Honest confidence, not a fake percentage.
AI-text detection does not work reliably, and no tool — including this one — can prove text is AI-written. OpenAI shut down its own detector for poor accuracy. A Stanford study found detectors wrongly flagged 61% of essays by non-native English writers as AI. Simple, formulaic, or technical human writing also trips these signals.
Treat everything here as evidence, not proof. Never use it to accuse a student, employee, or writer. Paraphrasing tools and newer AI models defeat all of these methods.
AI text is detectable because models fall into repeatable habits. The detector scores these directly.
Stock openers and filler like "in today's digital landscape," "delve into," and "it's important to note," counted directly against your text.
Words models reach for far more than people do, like leverage, robust, tapestry, and furthermore, measured per 1,000 words.
Human writing mixes short and long sentences. AI text clusters around one length. The tool measures this "burstiness."
A low type-token ratio, where the same words keep recycling, is a common signal of generated or padded text.
AI leans on the em-dash as an all-purpose connector. Overuse per 1,000 words is flagged.
Repetitive sentence openings, hedge stacking, false-balance framing, and canned wrap-up questions. These are the giveaways a reword alone won't fix.
An AI content detector is a tool that estimates how likely a piece of text was written or heavily assisted by an AI language model. This detector scores text on a 0–100 human-read scale by measuring the specific patterns AI writing tends to have: overused phrases and buzzwords, uniform sentence length, low word variety, decorative em-dashes, hedge stacking, and canned transitions. It reports the exact phrases and structural tells it finds rather than a single opaque percentage.
Yes. The AI detection and the one-click clean-up are completely free and unlimited, with no account and no sign-up. Everything runs in your browser.
No. This detector runs entirely in your browser using local text analysis. Your text is never uploaded to a server, never stored, and never used to train anything. This is different from most AI detectors, which send your text to a server to score it.
It runs a two-tier check. The first tier counts lexical tells: a list of banned AI phrases, buzzword density, type-token ratio for word variety, sentence-length burstiness, and em-dash overuse. The second tier looks for structural tells such as repetitive sentence openings, hedge stacking, false-balance framing, and canned wrap-up questions. Each signal that fails lowers the score from a starting 100, and the tool shows you exactly which ones fired.
No AI detector is 100% accurate, and this one does not claim to be. It is a diagnostic and editing aid: it surfaces the concrete AI-writing tells in a piece of text so you can judge and fix them. Short passages and heavily edited AI text are the hardest to classify, so treat the score as guidance, not proof.
Yes. Switch to "Scan a URL," enter a public page (a blog post, article, or landing page), and we fetch it, pull out the readable text, and drop it into the box so you can check it the same way. Pasted text never leaves your browser; in URL mode we fetch the public page you name and pull its text — we don't store the page or the URL. Pages that are mostly images or built entirely by JavaScript may not extract cleanly; paste the text if so.
Remove the flagged phrases and buzzwords, vary your sentence lengths so the rhythm is uneven, cut decorative em-dashes and canned transitions, and replace hedging with direct statements. The free one-click clean-up on this page does the mechanical phrase swaps for you; a full rewrite that preserves your meaning is available in the Anti-Slop Chrome extension.
Yes — it's free, unlimited, and needs no account or sign-up. You can detect AI writing online for free right here: paste text, get a score and the flagged tells, and use the rewrite, all without registering. There's no paywall and no "3 free checks" limit.
Two things. First, privacy: most detectors upload your text to their servers to score it; this one runs entirely in your browser, so your text never leaves your device. Second, transparency: instead of one opaque "% AI" number, it shows the exact phrases and structural tells it found and highlights the riskiest sentences, so you can judge and fix — not just take a verdict. It's honest that no detector, including this one, is reliable enough to accuse anyone with.
