You know, LLMs could do automated code reviews for each update to avoid things like this. It would be much better than unexamined updates.