Frequently asked questions
What is EyesOnText?
EyesOnText is an AI text detector that does not use machine learning to detect AI. It scores text using four statistical analyzers, readability variance, vocabulary diversity, sentence burstiness, and stylometric fingerprinting, each of which returns a named signal and threshold rather than an opaque probability.
How is this different from ChatGPT detectors like GPTZero or Copyleaks?
Those tools are trained neural classifiers: they compare text against patterns learned from labeled examples, and they cannot fully explain a given score. EyesOnText's analyzers are deterministic statistics and rules, so every flag traces back to a specific, inspectable measurement.
Does EyesOnText use AI to detect AI?
No. Detection is entirely rule-based: regexes, formulas, and statistical thresholds. Nothing in the detection pipeline is a trained model.
How accurate is it?
No detector, rule-based or machine-learned, is 100% accurate. Rule-based detection does not degrade as new language models are released, since it is not trained against any specific model's output, but it can still be wrong on unusual writing styles. We show the evidence behind every score so you can judge it yourself rather than trust a bare percentage.
Will it flag my human writing as AI?
Very formal, highly polished, or heavily edited human writing can occasionally score higher, which is why we ship genre presets (Academic, Marketing, and others) tuned to reduce false positives on those specific writing styles.
What does the correction feature do?
Correction (in Beta, on paid plans) runs flagged text through six steps: spelling, style, AI-specific phrases, passive voice, vocabulary repetition, and sentence-uniformity smoothing. Detection re-runs afterward, and any step that made the score worse is rolled back automatically.
Is the free preview really free?
Yes. No account is required. It is limited to one analysis per day per IP address and 500 words, for evaluation.
Do you store the text I submit?
Text submitted through the free anonymous preview is not stored, only the resulting score and verdict. Text submitted from a signed-in account is stored as a document in your account so you can review past analyses, and is deleted when you delete it or your account. Full detail is in our Privacy Policy.
Is there an API?
Yes, on Enterprise plans. It returns the same structured detection result as the web app: composite score, per-analyzer scores, and evidence.
Is there a Chrome extension?
Yes, free to install. It lets you right-click any selected text on any webpage to analyze it instantly.
Can I cancel anytime?
Yes, from account settings. Cancellation takes effect at the end of the current billing period, and nothing is deleted: you keep access to your existing documents at the free tier's limits.