Humans are fallible. No matter how well you review the code. It would be good if AI code can be identified. The kind of mistakes might be very different from the kind of mistakes human do. So when something happened, it can be identified to be AI generated. That data can be used to get an idea of what kind of mistakes are being committed.

It borks all the time but only because the disk becomes full due to snapper snapshots and pacman cache. I keep forgetting to clean up.