Those are exactly what you asked for. I would give it a 10 out of 10 for the prompt given.
If you interpret “spaces” as “only ASCII 32” then I would argue that you’re the one in the minority – in the Unicode world, there’s many encodings of the concept of “space” in text, and \s is the most robust encoding of that idea.
I’ve found that GPT-3.5 and GPT-4 only really understand Python and Javascript and Perl style “extended” regular expressions, though – trying to dumb it down to only be POSIX compatible regular expressions through prompting doesn’t work.