As an option, so it can automatically increment the brackets.
It's not though. It's literally asking the user "how old are you?" and not even caring if they lie. It's not even requiring a date, just a number of years.
I guess you haven't heard of Avenue Q?
Why would someone direct the output of an LLM to a terminal on its own machine like that? That just sounds like an invitation to an ordinary disaster with all the 'rm -rf' content on the Internet (aka training data). That still wouldn't be access on a second machine though, and also even if it could make a copy, it would be an exact copy, or an incomplete (broken) copy. There's no reasonable way it could 'mutate' and still work using terminal commands.
And to be a meme requires minds. There were no humans or other minds in my analogy. Nor in your question.
If you know that it's fancy autocomplete then why do you think it could "copy itself"?
The output of an LLM is a different thing from the model itself. The output is a stream of tokens. It doesn't have access to the file systems it runs on, and certainly not the LLM's own compiled binaries (or even less source code) - it doesn't have access to the LLM's weights either. (Of course it would hallucinate that it does if asked)
This is like worrying that the music coming from a player piano might copy itself to another piano.
That is a good point. Websites also if they are visited daily or if a web beacon or such can access the API.
Manually adjusting the brackets until 18+ (or just lying about the precise date) would grant more privacy. I can see making that trade-off though.