Post by @BoydStephenSmithJr@hachyderm.io

2026-07-22 15:58 UTC

I really hate how "LLM output is uncopyrightable" has become a meme. It's a woeful oversimplification. The U.S. copyright office says the necessary human creative effort for copyright can't come from LLM models or prompts, yes. But, they also establish that *in most cases* the model itself and it's outputs are derivative works of each and every piece of training material. That means they are absolutely subject to copyright protections, but based on the will of the rights holders of the training data, not the prompter. Fair use still applies, but is unlikely to cover anything other than research, and almost certainly doesn't cover commercial use. https://www.copyright.gov/ai/ -- Part 2 covers the "human creative effort" issues; Part 3 covers "training creates derivative works" issue.

Replies (1)

@BoydStephenSmithJr@hachyderm.io 2026-07-22 16:10

Supporting excerpts from the U.S. copyright guidelines (linked in parent post). "The training process [also] implicates the right of reproduction." -- Part 3, III, B. "outputs likely infringe the reproduction right and, to the extent they adapt the originals, the right to prepare derivative works. Some commenters noted that, depending on the content type and the audience, they may implicate the public display and public performance rights as well." --- Part 3, III, D "The Office concludes that, given current generally available technology, prompts alone do not provide sufficient human control to make users of an AI system the authors of the output." -- Part 2, II, D, 2 "Affirming the Office’s refusal, the court stated that 'copyright law protects only works of human creation'" -- Part 2, II, B "In most cases, however, humans will be involved in the creation process, and the work will be copyrightable to the extent that their contributions qualify as authorship." -- Part 2, II, B

Open ##4998051