“The fact that [AI]’s specific output cannot be predicted by users makes [AI] different for copyright purposes than other tools used by artists.”
They were specifically talking about the art-making AI, Midjourney in this ruling, but as a precedent, what does this mean for text and voice generating AI and the businesses that have been considering moving to them in place of hiring writers and actors?
That quote is the essential core of the ruling, AI tools are inherently different from tools like Photoshop, Illustrator, ProTools, Reaper, or even good old MS Word because the human is not directly in control of the output. Therefore, the output was not really the direct result of the marriage between a human’s thoughts, imagination, and emotion.
If BuzzFeed were using AI to write articles from completely from start to finish, would that be different from getting the AI to write a rough draft and then hand it off to a human to edit and rework?
And, since I work in audiobooks, where is the line in my industry?
If I record a book and then they use AI to fix errors instead of having me re-record sections and then having an engineer seamlessly blend those pickups into the original recording, the original recording does still originate with me — a human. Therefore, that would likely survive a copyright challenge.
But, if AI Enhanced Text-To-Speech (this is what we should all be calling audiobooks recorded purely by Artificial Intelligence, AI-TTS) is used to create the product from the very beginning, then through following the logic implied by this ruling, the final product cannot be awarded a Performance Copyright, fullstop.
To be clear, the words, the text of the book would still be under its original copyright, but since there was no human performer, there would be no performance copyright.
This should give the publishers barreling forward with a transition to AI pause. They will need to consider in what cases the performance copyright is the core of their protection and not the copyright on the original intellectual property.
Will the Internet become clogged by rearrangements of the audio found within an audiobook? Remixing them to tell different stories? To use the voices to make political statements those voices would never consent to make? To use those voices to create hate speech?
Maybe, just maybe, this ruling will slow down any AI-TTS plans that the publishers have been putting in place. And, since they will have likely already spent significant funds to leverage AI, let’s hope they consider putting that investment toward tools that will assist human narration instead of replacing it.