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.

https://www.theatlantic.com/technology/2026/05/pangram-ai-de...

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.