ChatGPT Detector estimates whether a passage was drafted by a GPT-family model. ChatGPT Detector scores predictability, uniform rhythm, and phrasing patterns associated with those models, then returns a probability for the GPT class. ChatGPT Detector highlights the sentences that most raised the score. ChatGPT Detector is aimed at that model family, so text from a different generator can score lower even when it is also machine-written. Editing by a person after generation pulls individual sentences toward the human side of the scale. A uniformity score rises when many sentences share both the same length and the same opening word choice.