I did some testing and found that Sapling is the least reliable of all the AI detectors on the list. I believe that false positives generated by it can lead to unfounded suspicions.
That's why we don't make the final conclusions based on any single one detector. Each can give a false positive. And that's why we are searching for multiple cases of AI usage, so that to be sure that the user is using AI for his posts and that's not an occasional false positive.