It has extremely low false positive rate in various tests (e. g. https://arxiv.org/html/2501.15654v1), I've never seen it mark something I knew was human written as AI.
Favorable benchmarks for Pangram exist, and it is one of the better AI detectors. But it's never been trustworthy as proof of authorship/misconduct, and it's getting less useful as models (and people's ability to steer them) improve.
At some point you're going to have to accept content is AI generated. There's heaps of stuff you probably read daily you don't even realise is AI generated.