Getty drops important copyright claims against stability AI, but the UK lawsuit continues
Getty Images removed the main claims of copyright infringement against stability AI in London’s High Court on Wednesday, narrowing one of the most common legal battles about how AI companies train models using copyrighted content.
This move doesn’t end the case completely – Getty still pursues other claims as well as other lawsuits in the US – but it highlights the grey area surrounding the future of ownership and usage of content in the age of generation AI. Development will also take place in just one day American judges sided with humanity Similar debates about whether to train AI in books without author’s permission violate copyright law.
Getty sued the startup AI behind the stable spread of AI image generators in January 2023 after claiming that stability used millions of copyright images to train AI models without permission.
Image Database Company also argued that many of the works produced by stable diffusion resemble copyrighted content used to train it. Some people even had a watermark, Getty said.
Both of these claims were removed Wednesday morning.
“The training claims may have been removed because Getty failed to establish sufficient links between infringement and UK jurisdiction over copyright law,” Ben Marin, partner at law firm EIP, said in an email. “On the other hand, the output request could have been probably removed (for example, by the photographer) because Getty failed to establish that what the model reproduced reflected a significant portion of what was created in the image.”
In the closing discussion of Getty, the company’s lawyers said they dropped these claims due to weak evidence and lack of knowledgeable witnesses from the stability AI. The company surrounded the move as strategic, allowing both it and the court to focus on what Getty believes to be a stronger and more winning claim.
What remains in Getty’s lawsuit are secondary infringement claims and trademark infringement claims.. Regarding claims for secondary infringement, Getty essentially argues that the AI models themselves could infringe copyright laws and that using these models in the UK allows for the import of infringement articles, even if training was done outside the UK.
“Second-infringement is through the model itself, through the broadest Genai companies that train outside the UK and through the model itself, which could be “infringing articles” that have since been imported into the UK,” Maling said.
A Stability AI spokesperson told TechCrunch that the startup is “pleasant to see Getty’s decision to withdraw multiple claims after the testimony is over.”
The spokesman also said that stability is certain that the request to pass on Getty’s trademark is a failure as consumers do not interpret the transparent layer as a commercial message from Stability AI.
Getty’s US division sued stability AI in February 2023 for trademark and copyright infringement. In that case, Getty claimed that stability used up to 12 million copyright images to train the AI model without permission. The company is seeking damages for 11,383 work, $150,000 per infringement, totaling $1.7 billion.
Separately, another complaint has been named alongside Midjourney and Deviantart after a group of visual artists sued the three companies for copyright infringement.
Getty Images has its own Generated AI Offering Getty Istock stock photos and video library-trained AI models. This tool allows users to generate new licenseable images and artwork.