"Our classifier is not fully reliable. In our evaluations on a “challenge set” of English texts, our classifier correctly identifies 26% of AI-written text (true positives) as “likely AI-written,” while incorrectly labeling human-written text as AI-written 9% of the time (false positives). Our classifier’s reliability typically improves as the length of the input text increases."
https://openai.com/blog/new-ai-classifier-for-indicating-ai-...
Utterly useless for teachers. You can't go around accusing 1 in 11 innocent students of plagiarism, and over the course of a semester you'd have virtually every student get falsely accused at least once. They may as well have not released this tool at all.
People claiming to be able to reliably detect LLM generated content have been found to be selling snake oil.