This article is more than 1 year old

Judge lets art trio take another crack at suing AI devs over copyright

Just the facts, ma'am. More facts, in fact

A US federal judge almost threw out a lawsuit brought by artists accusing text-to-image AI developers of copyright infringement, but decided to give the creatives a chance to improve and resubmit their complaint.

In January, illustrators Sarah Andersen, Kelly McKernan, and Karla Ortiz sued Stability AI, Midjourney, and DeviantArt, and sought class-action status. They argued the developers had unlawfully scraped billions of images from the internet – including their copyrighted art – to train AI models to generate digital images from users' natural language descriptions.

People can use tools like Stability's DreamStudio or Stable Diffusion to create images that mimic people's individual styles – effectively ripping off their work and endangering their livelihoods, the artists claimed. You can ask certain models things like: "Draw a comic about a recruiter interviewing someone for a programming job, in the style of Sarah Andersen" and they will output pictures matching that query in something approximating her unique style.

"These resulting derived images compete in the marketplace with the original images. Until now, when a purchaser seeks a new image 'in the style' of a given artist, they must pay to commission or license an original image from that artist," the trio's complaint [PDF] read. 

"Now, those purchasers can use the artist's works contained in Stable Diffusion along with the artist's name to generate new works in the artist's style without compensating the artist at all."

The artists, represented by attorney Matthew Butterick and the Joseph Saveri Law Firm, argued that each image generated by AI software was created by combining relevant images scraped from its training dataset. And since those original works are protected by copyright, they argued, all AI-generated knock-offs infringe upon the artists' rights.

Stability, Midjourney, and DeviantArt disagreed, and filed motions to dismiss the claims. Federal district Judge William Orrick, sitting in San Francisco, California, said in a hearing this week he was leaning towards binning the lawsuit – but is, instead, going to allow the plaintiffs to refile a fresh complaint.

He pressed them to "provide more facts" to back up their claims and show that their own artwork had been directly involved in helping create the various tools.

"Otherwise, it seems implausible that their works are involved," he was reported as saying.

Judge Orrick warned them that their claims – that the fact AI could generate images based on text descriptions containing their names violated copyright laws – are unlikely to hold up in court on their own.

"I don't think the claim regarding output images is plausible at the moment, because there's no substantial similarity between images created by the artists and the AI systems," he said.

Whether content generated by AI systems is considered fair use or not is a very open question right now. Authors have similarly accused companies developing AI tools like those of OpenAI of violating copyright laws by scraping their books without permission. 

The artists' lawyer told The Register his team will amend their complaint. 

"At the hearing, Judge Orrick indicated that plaintiff Sarah Andersen's direct infringement copyright claim against Stability AI could go forward. He indicated that he would give the plaintiffs permission to amend their copyright infringement claims against Midjourney and DeviantArt," Butterick said.

"It is not at all unusual at this stage of the litigation for the judge to give plaintiffs permission to amend their complaint, and for the plaintiffs to do so. This also happened in our case challenging GitHub Copilot."

The Register has asked Stability, Midjourney, and DeviantArt for comment. Stability declined to speak on the matter. ®

More about

TIP US OFF

Send us news


Other stories you might like