I have also noticed that and I think that the text may be written by a human and has been polished with the help of AI tools. If someone tries to fix the grammar issues of a human written text with the help of those AI based software then the detectors will still detect that text as AI generated text and that's why I think moderators allow such posts because they know that thing. I think one should avoid using those AI based software at all cost because those tools can fix the grammar, but they also make your hard-work as a waste of time.