AI in climbing

Drew, you may own the photograph, but any derived or derivative works produced with an LLM cannot be copyrighted. This is a major legal issue that everyone seems to just sluff off.

AI is not a universal copyright remover if sprinkled on. Drew says he took the photographs and wrote the route descriptions, which would both be copyrighted. Using AI to generate an HTML app to present them would not impair those copyrights. As background, in the U.S. the Copyright Office has stated they do not think AI requires a new legal framework for derivative works. What matters is where the human original expression is embodied.

As a practical matter, presentation code does not really matter much for copyright in general. If you put your text and photos into a Microsoft Word file, you don't gain ownership of the XML formatting code that Word generates. If you run a Wordpress website, most of the front-end formatting code is probably generated by Wordpress or plugins, so you would not hold copyright on that. That doesn't grant anyone else any rights to the original copyrighted content you placed inside that formatting, though.

The Copyright Office did say they will not grant copyright on wholly generated works. So if someone starts from a blank screen and prompts "make a photo of El Cap with all the climbing routes traced on it," or "write a route description for Mescalito on El Cap," the prompting person would not be able to register a copyright on that output.