Generative Filled Art Probably Isn’t Copyrightable
We should look at what the United States Copyright Office (USCO) thinks. Unfortunately, it wasn’t willing to comment about specific new technologies and that makes sense — things are moving too quickly for it to spend time specifically addressing each new AI technology.
“The Copyright Office declines to comment on any specific AI technologies. We would refer you to our recent policy statement on copyright registration for works containing components generated by AI, as well as our recent listening session on the impact of AI on visual arts works,” the Office tells PetaPixel.
That policy statement actually points to a case PetaPixel covered earlier this year, where the Office rescinded a copyright it had granted to an AI-generated comic.
“The Office reviewed a registration for a work containing human-authored elements combined with AI-generated images,” the Office says.
“The Office concluded that a graphic novel comprised of human-authored text combined with images generated by the AI service Midjourney constituted a copyrightable work, but that the individual images themselves could not be protected by copyright,” the Office says.
“In the Office’s view, it is well-established that copyright can protect only material that is the product of human creativity.”
Basically, while the images did have human input, they were entirely AI-generated and thus were not copyrightable. The only copyrightable material was the text, which did not involve AI at all.
“Under current copyright office guidelines, the portions of an image that were created with generative AI technologies would need to be disclosed and disclaimed as part of the registration process,” Maddrey says.
On that note, the USCO has said that, “for example, when an AI technology receives solely a prompt from a human and produces complex written, visual, or musical works in response, the ‘traditional elements of authorship’ are determined and executed by the technology—not the human user… As a result, that material is not protected by copyright and must be disclaimed in a registration application.”
Luckily, it does seem like there is some level of certainty around this one aspect of Generative Fill. Unfortunately, it doesn’t stay that way.
‘New’ Art Made with Copyrighted Images is Probably Protected
The second main question on this topic is whether or not you as a photographer can legally stop someone from using your work as the basis of new work that is expanded using Generative Fill.
On the surface, the answer seems like it will be “yes,” but unfortunately it’s not cut and dry.
“We would argue that the addition of material in this way would not constitute a new work such that the underlying copyright protections could be stripped from the original author. The USCO won’t accept it as work subject to registration,” Maddrey says.
Maddrey adds that the position of the ASMP would be yes, a photographer would have legal standing to go after someone who was using their work without permission to create new work using AI. However…
“Of course [the AI user] then could raise the affirmative defense of fair use, but that opens up a whole other can of worms that the courts are just now starting to implement in light of Warhol v. Goldsmith.”
And that is the sticking point: would it qualify as fair use? The results of the recently decided Warhol v. Goldsmith case seems to indicate that expanding an image using AI wouldn’t be enough to constitute fair use, since the Supreme Court chose to eliminate the oft-used point raised in fair use analysis: simply adding “some new expression, meaning or message” does not, in and of itself, create a transformational use.
But does that wholly close the book on this specific type of generative AI? As Maddrey notes, we wouldn’t likely know until a case went to court. Photographers are probably able to protect their copyrighted photos against this kind of use, but that’s the key word — probably.
