On Friday three artists filed a proposed a class action suit against major AI image generating companies – Stability AI, Deviant Art and Midjourney – alleging they infringed on copyright laws through the use of collage tool Stability Diffusion.
“Stable Diffusion contains unauthorized copies of millions – and possibly billions – of copyrighted images. These copies were made without the knowledge or consent of the artists,” their lawyer, Matthew Butterick, alleged on the litigation website on Friday.
Stability AI released text-to-image tool Stable Diffusion in August 2022. It was trained on pairs of images and captions scraped from the web. It stores compressed copies of the training images and uses them to recombine the images into new “collage” artwork based on the existing artist-made image.
Butterick alleged that whether the final images resemble the training image does not matter as it competes with the originals in the marketplace.
“At minimum, Stable Diffusion’s ability to flood the market with an essentially unlimited number of infringing images will inflict permanent damage on the market for art and artists,” said the legal eagle before pointing out Stable Diffusion is not licensed and calling it a “parasite.”
Defendant DeviantArt, a web-based artist community since 2000, has released a paid app built using Stable Diffusion called DreamUp, which Butterick says crowds out the work of human artists with a flood of AI-generated art. Meanwhile, Midjourney offers a text-to-image generator trained on the scraped images.
Within the complaint [PDF] are allegations of both vicarious and direct copyright infringement, violations of the Digital Millennium Copyright Act (DMCA), violations of the rights of publicity, violations of California’s unfair competition law, and breach of contract for DeviantArt’s Terms of Service.
The three plaintiffs in the case include artists Sarah Andersen of webcomic Sarah’s Scribbles, Kelly McKernan and Karla Ortiz, who are seeking compensation and injunctions on behalf of all artists affected by Stable Diffusion.
“Humans cannot help bringing their own humanity into art. Art is deeply personal, and AI had just erased the humanity from it by reducing my life’s work to an algorithm,” said Andersen recently in an op-ed for the New York Times.
Within the article, she described both how her work had been appropriated for use among the alt-right and the experience of her style’s ubiquity in the realm of AI generative art.
“Once the features that we consider personal and unique – our facial structure, our handwriting, the way we draw – can be programmed and contorted at the click of a mouse, the possibilities for violations are endless,” said Andersen.
Fellow defendant McKernan tweeted that her name had been used and profited from at least 12,000 times before sharing her Venmo code, just in case.
For the AI “artists” profiting on my name, share some of that wealth. I have bills and rent to pay, groceries to buy, and a kid to support. The least you can do is throw $20 my way. I’m a real human artist barely making ends meet and your actions have real consequences. pic.twitter.com/5NnKchgMIk
— Kelly McKernan (@Kelly_McKernan) January 15, 2023
Individuals harmed or banding against AI copyright infringement are plentiful, but so are those advocating to keep the laws surrounding AI art minimal.
“Tbh it’s not that (digital) artists will be replaced by AI. Digital artists who use AI will outcompete those that don’t. This is only in regular output for work as raises base level. Non-digital art forms were not killed by digital art, plenty of space in our society for all,” said Stability AI CEO Emad Mostaque in a tweet last week.
A website calling the lawsuit “frivolous” created by anonymous authors calling themselves “tech enthusiasts” has since appeared as a response, laying out both Butterick’s original announcement and a response side by side.
The rebuttal claims that “the rights of creators are not unlimited,” and that all must roll with the times and new technologies. The authors also accuse Butterick of hypocrisy for using images from research papers without compensation or consent, although the lawyer very clearly gives citation for the work, making it clear who published the original work.
Others have expressed worry that big artists will be able to reduce the tool’s availability to lesser known ones.
One Twitter user pointed out that in US law, scraping online images for datasets can fall under the fair use copyright doctrine – although what the result could hinge on is whether the artwork could ultimately be deemed “not transformative,” and whether it has an effect on the market for the original work, which the lawsuit appears to be alleging that it does.
that “not transformative” bit is really interesting, bc whether or not a work is “transformative” is at the crux of fair use — a concept in US copyright law that lets copyright-protected works be used, sans license, for certain things.
This is not Butterick’s first rodeo when it comes to tackling AI copyright intrusion. In November he was a part of a team that filed a lawsuit against GitHub Copilot for its alleged “unprecedented open source software piracy.”
That lawsuit is still in progress. ®