do LLM models make accidents?

i think many times we look at things like this happening and think "can never be me", but it really could be me next time.

Someone recommended this in the replies: https://github.com/Dicklesworthstone/destructive_command_gua...

Any other must-have tools like this?

Oh no. Someone gave an AI access to delete all the files on their system and then an AI deleted all the files on their system.

And then that same person basically said, "I'll try a different model."

Lesson clearly not learned.

I made the same mistake once with a junior dev