Frequently asked questions
How does the free AI content detector work?
Paste text (100+ words) or enter a URL. The detector runs a statistical stylometric analysis: it checks for AI-typical marketing phrases, measures vocabulary diversity, analyzes sentence-length rhythm (burstiness), looks for repeated formulaic openers, and counts punctuation signatures like em-dash density. Each signal is shown to you with its exact evidence — nothing is hidden in a black box.
Can AI detectors be wrong?
Yes. All AI detectors, including this one, are heuristic — they measure statistical patterns, not authorship. False positives happen: formulaic human marketing copy, translated text, or heavily edited prose can trigger AI signals, and skilled human editing can mask them. Never use any detector score as the sole proof in academic, legal, or employment decisions. Treat it as one data point alongside provenance, revision history, and human review.
What does the 0–100 AI-likelihood score mean?
Under 35 means the writing patterns look likely human; 35–64 means the signals are genuinely mixed; 65 and above means multiple independent AI-typical signals fired together. Scores near the middle are common for edited or hybrid human-plus-AI text — that ambiguity is honest, not a bug.
How is this different from other AI detectors?
Most detectors return a single number with no explanation. This tool shows you every signal that fired — the exact phrases matched, the measured vocabulary ratio, the sentence-rhythm values — so you can judge the evidence yourself. It also states its limitations openly instead of pretending to certainty no statistical tool can offer.
Why does the detector need at least 100 words?
Stylometric signals need enough text to be meaningful. Vocabulary diversity, sentence rhythm, and repeated patterns cannot be measured reliably on a paragraph — short samples produce noise, not evidence. One hundred words is the minimum for the statistics to stabilize; longer text gives more reliable results.