If most languages have a builtin, then the challenge is too simple
There are some exceptions, for example I like the existence of the Add Two Numbers challenge as a "language showcase", but they are exceedingly rare. In other words, if all the answers are just "use builtin" then the challenge won't be saved by banning builtins.
The fact that a language has a weird builtin can itself be interesting
An answer that sometimes gets brought up in discussions about builtins is this one, which uses Mathematica's built in "Does the image contain an upright goat" function. As can be seen, many people like that answer due to the sheer absurdity of a programming language with that builtin. I don't think anybody would accuse that challenge of being "too simple", though some don't agree with me that the linked answer is "interesting".
Another example is this one, which is interesting (to me) because it relies on less-than-obvious language behavior to use a single general-purpose function to solve a much more specific challenge. Again, not everybody finds such answers interesting, but there would be very few answers on the site at all if we only accept answers that everybody finds interesting.
A challenge that's uninteresting in some languages can be interesting in others
So you make a challenge that isn't trivial, the logic is pretty interesting, but the result is "useful" so several languages have a builtin for it. So what? It may not be an ideal situation where every answer is great, but that doesn't mean you should shove the challenge in the horrendous mess that is "trying to ban builtins" just to reduce the number of answers in hopes that the remaining ones are interesting. It's just not worth it.