LONDON – AI Firm Stability AI Faces Off Against Getty Images

Stability AI, an artificial intelligence company, came close to securing a favorable outcome in a legal dispute over intellectual property with Getty Images on Tuesday.

Getty Images, based in Seattle and the owner of a vast online collection of images and videos, initiated a high-profile lawsuit against Stability AI that was heard in the UK High Court back in June.

This case is part of a larger trend, with various lawsuits sprouting up from film studios, artists, and writers who are questioning how tech companies are using their work to inform AI chatbots.

In the ruling released on Tuesday, Getty managed to win on a trademark issue but lost on a secondary copyright claim. Interestingly, both parties declared some form of victory.

Getty Images touted this as a significant win for those holding intellectual property rights. They said it in a statement that was quite adamant about its implications.

However, despite the positive news for Getty, its stock dipped by 3% before the market opened in the U.S.

On the other hand, representatives from Stability expressed satisfaction with the outcome. Christian Dowell, General Counsel at Stability, mentioned that this decision addresses the key copyright issues at stake.

Getty accused Stability of committing a “brazen breach” against its image library, asserting that the development of Stability’s AI image generator, known as Stable Diffusion, was extensive.

Although Getty initially made claims regarding copyright infringement, they withdrew the main copyright accusations during the trial, suggesting doubts about their success in those arguments.

The company also alleged trademark infringement, claiming that some images produced by Stability’s AI included Getty watermarks.

In her ruling, Judge Joanna Smith acknowledged that Getty’s trademark claims had “partial” success but emphasized that her conclusions were limited in nature.

Stability contended that the case wasn’t right for the UK courts, arguing that the training of the AI model occurred elsewhere, notably on Amazon’s servers in the United States. They also mentioned that only a small percentage of the outputs from the AI closely mirrored Getty’s work.

Tech companies maintain that legal principles like “fair use” in both the U.S. and the U.K. permit them to utilize large volumes of text and images for training their AI systems.

Beyond this, Getty continues to pursue a claim of “secondary infringement,” asserting that even if the training happened elsewhere, offering Stability’s services to users in the UK equated to introducing illegal copies into the country.

Judge Smith dismissed these claims, ruling that Stable Diffusion’s AI does not infringe on any copyrights since it hasn’t stored or reproduced any copyrighted material.

Getty is also involved in a copyright infringement lawsuit against Stability in the United States, having initially filed in 2023 and then refiling in San Francisco federal court in August.

This legal confrontation underscores the broader tensions arising from the surge of generative AI technology, creating friction between tech firms and creative sectors.

Notably, Anthropic has agreed to a $1.5 billion settlement related to allegations that they used pirated works to train their Claude chatbot.

In another case, a judge dismissed a lawsuit involving 13 authors who had similar claims against Meta Platforms, which is developing its AI system, Llama.

Moreover, Warner Bros. has taken legal action against Midjourney for copyright infringement, arguing that its image generator allows users to create AI-generated visuals of copyrighted characters. Disney and Universal have also filed joint lawsuits against Midjourney, alleging it has unlawfully used their libraries to generate unauthorized images of iconic characters.