That is the weird part of this all AI coding, you would think that was the case, but in reality, you found all kind of architectural gaps due to lack of constraints enforcement. The article describes in more details about this high believability software phenomenon.
Interesting, nice article.