There are three things to unpack there:

Tools don't create art, neural networks wielding those tools create art.

Right now, human NNs are the most complex around the block, so our anthropocentric egotism tries to gatekeep art to humans... ignoring all the animal art out there, like for example birds building "beautiful" nests to attract mates (beautiful to each other, not necessarily to humans), all the art going on between fish, cephalopods, dolphins, whale songs, etc. There is also no guarantee that human NNs will remain supreme forever... and what then, will humans stop creating art, or will the ant tell the elephant that its art is not a thing?

Tools DO use existing human work, otherwise city photography could never be art, cultural photography could not be art, definitely a Campbell soup can could never be art... and so on. The Camera obscura has been used to "cheat" at art since possibly the paleolithic, then extensively "abused" by the likes of Leonardo da Vinci to copy both natural and human works.

Modern AI does way more than "copying", it abstracts the underlying patterns, then integrates those abstractions with a prompt, to "make up" an output. Sometimes the output of the abstraction of an "A" looks like an "A", other times it doesn't. People keep putting AI down for "hallucinating"... but you can't claim that it "hallucinates" and "copies" in the same sentence.

For an intro on how modern AIs work, I'd suggest checking: Neural Networks, by 3Blue1Brown

AIs have not been "copying" for several decades already, modern AIs are even farther away from that, and it's just the tip of the iceberg.