Unfortunately, only a single PR like this comes to mind. Most AI authors we've seen were identifiable mainly by overly verbose PR descriptions, meaningless code changes and copy-pasting more AI output when questioned.