Which ChatGPT detector is most accurate?

Question

Accepted Answer

No detection device can be perfect for all situations, so any device that markets itself with a single accuracy score is talking about only one testing situation and not yours. Accuracy depends on conditions. Detection devices perform the best with raw, unedited ChatGPT output of 300 words and more and the worst with very brief, heavily edited or non-native English writing. Scores disclosed by the manufacturers in comparison to different materials cannot be applied literally.
The proper way to compare is to assess reliability. What matters most is the number of false positives: the case of a real human writing coming up as a false positive. Use devices that provide data on likelihood of their judgment, so that the users can understand why.