What our checker actually measures
The free ClearCopy AI text checker looks for assistant-style opening and closing phrases, repeated formulaic transitions, repeated sentence openings, unusually consistent sentence lengths, and repeated five-word passages. These are ordinary writing features. Humans and AI systems can produce every one of them.
What the results mean
The results identify specific phrases and patterns worth editing. They do not give a calibrated probability, determine whether ChatGPT wrote the draft, or assess the legal or academic provenance of the content. A negative result does not establish human authorship.
Why we do not give an AI percentage
Reliable authorship classification needs a trained model and evaluation on representative human and AI writing, with measured false-positive and false-negative rates. ClearCopy does not have such a model. A percentage would misrepresent what the algorithm knows.
Even providers of trained models warn of false positives and limitations for short or unusual text. Turnitin's published AI Writing Report guidance cautions against using the indicator as the sole basis for adverse action and sets length requirements for a score. OpenAI discontinued its own AI text classifier in 2023 because of low accuracy.
Different from a statistical watermark check
OpenAI's textGrain is a statistical word-choice watermark. Our pattern checker cannot determine whether textGrain is present or absent and cannot validate watermark removal. See watermarks explained.
Privacy and responsible use
The browser checks the text on your device. Nothing is uploaded to an AI-detection API. Use the results to improve clarity and spot repetitive patterns, not to accuse someone of misrepresenting authorship. Follow rules that require disclosure of AI assistance.