I understand the issues related to LLM leaking and re-distributing "private" information, but I'm curious which category of concerns you're referring to. Would you mind giving some context (genuinely curious) ?

You can look at the vulnerabilities found graph of github. It stays about the same for years and then skyrockets up at around the time LLMs were invented.

And they are all pretty simple vulnerabilities, exploitable even to people knowing nothing about how to get root access from a binary that has an out of bounds condition somewhere in randomly shuffled memory layout in a specific version if a C program.