In January, three visual artists filed a class-action copyright lawsuit against Stability AI, the startup that created Stable Diffusion. In February, the image-licensing giant Getty filed a lawsuit of its own.
“Stability AI has copied more than 12 million photographs from Getty Images’ collection, along with the associated captions and metadata, without permission from or compensation to Getty Images,” Getty wrote in its lawsuit.
Legal experts tell me that these are uncharted legal waters.
“I’m more unsettled than I’ve ever been about whether training is fair use in cases where AIs are producing outputs that could compete with the input they were trained on,” Cornell legal scholar James Grimmelmann told me.
Generative AI is such a new technology that the courts have never ruled on its copyright implications. There are some strong arguments that copyright’s fair use doctrine allows Stability AI to use the images. But there are also strong arguments on the other side. There’s a real possibility that the courts could decide that Stability AI violated copyright law on a massive scale.
That would be a legal earthquake for this still-nascent industry. Building cutting-edge generative AI would require getting licenses from thousands—perhaps even millions—of copyright holders. The process would likely be so slow and expensive that only a handful of large companies could afford to do it. Even then, the resulting models likely wouldn’t be as good. And smaller companies might be locked out of the industry altogether.
And, of course, if the courts don’t rule in favour of the copyright holders, it will make IP worthless. You will no longer have any rights to your own work. Disney would have exclusive rights to images of Mickey Mouse. Musicians could no longer on their own music. Very scary times.
I have a different preliminary view of AI-generated imagery and copyrights.
When AI draws an image of Mickey Mouse or anything else that’s trademarked, copyrighted, or patented, that imagery is still covered by the same regulations as if I had drawn it myself.
However, when AI learns from looking at tens of thousands of images of Mickey Mouse, other cartoon figures, real mice, various rodents, and different drawing styles and somehow incorporates generalized bits and pieces of what it’s learned to create something new, that process seems to fall within existing fair use parameters. For that matter, it’s precisely what we do. We learn from everything we experience. We store bits and pieces of what we’ve learned, then use that information to create something different.
I don’t see how the AI software companies could be prevented from analyzing copyrighted imagery any more than Google could be precluded from analyzing that same information to incorporate into its search results. The onus still falls on the end user to use that information in ways that are consistent with applicable laws.
At most, I can imagine the courts deciding that the AI image-generation tools can’t use what they learn to recreate and distribute imagery to users that a reasonable person would conclude infringes on the legal rights of others.
From what I’ve read, this is already the direction Adobe is heading with their Firefly AI. Adobe is restricting Firefly’s learning to the imagery that Adobe has rights to use. They also plan to limit the imagery it creates to avoid copying other identifiable artists’ and photographers’ styles.
I asked Firefly to “Create an image of Mickey Mouse standing in front of a McDonald’s sign.” Here’s what it came up with. The AI was obviously familiar with Mickey Mouse and McDonald’s, but it created something that doesn’t violate any intellectual property laws I can identify.
From what I read, Adobe is also using copyrighted images where free-to-use Wikimedia Commons permission exists and images from its stock service, where there’s a clause in the contract that gives Adobe permission to use the images. I’d need to check more into this, though.
I wonder how it will work going forward? All that is fine for existing copyright material. What happens if someone used the technology to create a new character / any original work? Does that become copyright of the artist, or of the tech that it could be claimed created it. Will copyright be volition or execution? It seems there’s a minefield of legal conundrums on the way. A least the lawyers will get rich off the back of it – who’d have thought it!!
Unless, of course, AI-legal determines the outcome.
Yes, I agree, along with all kinds of societal disruptions.
The short-term effects seem manageable to me. But stretch AI improvements out over the next ten or twenty years, and I envision changes that will equal or exceed the changes caused by the industrial revolution of 200–250 years ago, for better or worse.