Stability AI’s Getty Win Puts Copyright Law in Question
▼ Summary
– Stability AI largely won against Getty Images in a UK legal battle over training AI models, though the case failed to set a clear precedent on using copyrighted works for training.
– Getty Images dropped its core claim about training on copyrighted material mid-trial due to weak evidence, weakening the case’s impact on AI copyright law.
– The UK High Court ruled that Stability AI infringed Getty’s trademark by generating images with its watermarks but rejected the secondary copyright infringement claim.
– Getty Images is pursuing a separate case against Stability AI in the US, having refiled it in California after initially filing in Delaware.
– This case is part of broader copyright disputes between AI companies and creative firms, with recent settlements like Anthropic’s $1.5 billion payment to authors.
A recent legal decision in the United Kingdom involving Stability AI and Getty Images has left fundamental questions about artificial intelligence and copyright unresolved. While the London-based firm secured a favorable outcome on the primary copyright claim, the ruling failed to establish a clear legal precedent regarding the use of copyrighted materials to train AI systems. This leaves both technology developers and content creators in a state of uncertainty, awaiting future cases to provide more definitive guidance.
The lawsuit, initially filed in 2023, represented the first major AI copyright dispute to reach England’s High Court. Getty Images had accused Stability AI of unlawfully scraping millions of its copyrighted photographs and videos to train the popular Stable Diffusion model. However, as the trial progressed, Getty chose to abandon its central argument concerning the training data itself. Legal experts suggest this move was likely due to challenges in presenting sufficiently compelling evidence to support their claim.
In her written judgment, High Court Judge Joanna Smith did rule that Stability AI had infringed upon Getty’s trademark rights. The court found that the AI model had generated images containing distorted versions of the Getty Images watermark. Conversely, the judge dismissed Getty’s allegation of secondary copyright infringement. She determined that because Stable Diffusion does not store or reproduce the actual copyrighted images, but rather learns from them to create entirely new content, this specific claim could not stand.
Getty Images is now focusing its legal efforts on a parallel case underway in the United States. The company first initiated proceedings in Delaware federal court shortly after filing the UK lawsuit. In a strategic shift this past August, Getty voluntarily dismissed that initial American complaint and promptly refiled it in a California court. This maneuver is widely interpreted as an attempt to secure a more favorable legal interpretation in a different jurisdiction.
This dispute is merely one thread in a rapidly expanding tapestry of litigation concerning generative AI. The core conflict revolves around whether AI companies require permission, and potentially must provide compensation, when their models learn from copyrighted books, artwork, and music. In a related development, AI firm Anthropic recently agreed to a settlement reportedly valued at $1.5 billion to resolve a lawsuit brought by a consortium of authors. In another significant move, Universal Music Group dropped its copyright claims against AI startup Udio. This decision was part of a broader strategic partnership to collaboratively develop a new AI-powered music creation platform, illustrating how some disputes are evolving from pure litigation toward commercial collaboration.
(Source: The Verge)
