@darius What about the direction the copyright office is taking where the output of AI is not eligible for copyright? https://www.msn.com/en-us/news/politics/us-copyright-office-rules-ai-generated-artwork-content-not-legally-protected/ar-AA17TMuX

Calling these outputs the speech of the person/company who coded the AI would seem to imply that SOMEONE should be able to copyright these outputs.

Personally I prefer the idea that all these outputs are automatically public domain, given that they didn't get permission for most/any of the code/images/text in the training corpus.

But if strict liability hamstrings these LLMs, that works too.