After months of pushback and criticism from artists in a wide swath of industries, Stability AI, the creators of the popular AI art tool Stable Diffusion, have been hit with a pair of lawsuits over copyright violations that could change how the company does business. Most notably, photography giant Getty Images announced this week that it started legal proceedings alleging “Stability AI unlawfully copied and processed millions of images protected by copyright.”
Stability AI was also named late last week in a California class action suit that alleges “direct copyright infringement, vicarious copyright infringement related to forgeries, violations of the Digital Millennium Copyright Act (DMCA), violation of class members’ rights of publicity, breach of contract related to the DeviantArt Terms of Service, and various violations of California’s unfair competition laws.”
Late last year, artist Darek Zabrocki, whose resume includes Planet of the Apes, Netflix’s Love + Robots, Sonic The Hedgehog 2 and major video game franchises like Call of Duty: Modern Warfare, Assassin’s Creed and The Witcher, found his artwork was being used as well. Zabrocki told Paste that he has always been excited about incorporating new technologies into his work. But, while the tech behind Stable Diffusion looks promising, the underlying ethics are completely off.
“AI generators are trained on human art and human creations. It takes artists’ work without their consent in order to make ‘new’ pictures or to ‘copy’ a style of an artist to feature that on an AI-generated picture,” Zabrocki told Paste via email. “Taking a chunk of artists’ work, making a mish-mash of these art pieces in order to spit out a new piece within a minute is what makes AI problematic, and it seriously violates the copyright laws on many levels. Especially if such ‘art’ is going to be used commercially.”
In September 2022, a company known as Spawning created the site HaveIBeenTrained.com to search the LAION-5B image set, a collection of 5.8 billion images that have been used to train popular AI art models. Shortly thereafter, ArsTechnica reported that a California-based AI artist who goes by the name Lapine discovered private medical record photos taken by her doctor in 2013 referenced in the LAION-5B set. In December, Stability AI announced it would allow artists to remove their work from the training dataset for an upcoming Stable Diffusion 3.0 release.
As a concept artist, Zabrocki said he has always been excited about incorporating new technologies into his workflow. Years back, it was all about implementing 3D, then VR. “Some new tools become standard, some of them drop off never to be heard from again,” he said. Zabrocki recalled testing Midjourney during its first phase of development and it “seemed like a cool way to generate random shapes and rough compositions.”
“In order to produce art on the top level in the entertainment industry, you must be flexible and keep your mind open to new things,” he said. “The tech behind AI itself looks and sounds promising, boosting up your efficiency by automating some of the process or helping you during tedious tasks while you’re constantly working under the time pressure. Of course you can still spot some weird problems with rendering or some repeated glitches but I think it’s only getting better now. I cannot say I would just do 100% of my commercial work with AI-generated pieces though. I still do think that for brainstorming and ideas, it’s cool, but the ethical side needs to be resolved and formulated by the law.”
Moving forward, Zabrocki said he wants to see clearly stated regulations and boundaries on AI’s use of art. But he doesn’t want a ban. In his words, people need to figure out how “AI can help artists, not blatantly push to replace them.”
Jimmy Gunawan is a generalist/technical CG (Computer Graphics) artist currently based in Sydney, Australia, and created Blender Sushi, a blog dedicated to exploring 3D art. Gunawan said the main lesson as computer artists is “that you must continuously care about your tools, apart from having to also care about the art.”
Gunawan said he’s only been using AI Art tools for a few months, but he’s been fascinated with its progress over the past year. In addition to Midjourney, Gunawan said he dabbled with DallE Craiyon and Stable Diffusion, starting in their early days. An avid Blender 3D user, Gunawan said he also uses add-ons that utilize AI, such as Dream Textures.
Gundawan points out the tech has gotten so popular, you can use your iPad or iPhone to experiment with AI Art. Gundawan said to protect works from “AI Artist Thieves”, he believes artists really need to make a name for themselves and have some kind of presence online, via YouTube, Tiktok or even Twitter.
“I believe the AI tool has an amazing potential for every individual artist to let them be more open with variations of ideas, open up their mind on ‘remixing’ and diffusing ideas with others. The AI can think and add random variations a lot faster than our brain,” he said. “Although of course, in the end, the artist still has to do some filtering and finding the one that really sparks and enhances the original ideas.”