In my case it is rarely that I use LLM to write comments but rather I frequently use an LLM on my finished comment to fix things I miss as a non native speaker.

The content of the comment is my unique opinion and my unique writing and I mostly also make sure to remove stupid things like directional quotation marks.

But yes, it is possible to be very much human but also trigger certain peoples AI detectors.

What makes you think it's LLM generated?