Is it less secure if I simply have nothing to do with language models whatsoever? I take that to be the parents point, but I know you are correct and yes, user restrictions should be a necessary part of the way people who use LLMs setup their system to use them in a safe manner.