I like the faith you have that people weren't making null-pointer mistakes before LLMs.

It did happen before LLMs, but there are well documented process to catch them. Google literally wrote the book on SRE best practices.

error handling is very basic, the only explanation these kind of bad code to get pushed to prod is LLMs and high trust on LLM automation.

they wont admit this publicly anyway, there is too much money invested on LLMs.