Getty Images has withdrawn its primary copyright infringement claims against Stability AI at London’s High Court, dramatically narrowing an ongoing and closely watched legal dispute surrounding the use of copyrighted content in AI model training.
The decision does not mark the end of the lawsuit altogether. Getty continues to pursue other legal allegations, including secondary infringement and trademark infringement, both part of a broader case which includes separate ongoing litigation in the United States. The shift comes just a day after an influential ruling in the U.S. sided with the AI company Anthropic in a case concerning similar copyright questions—specifically, the training of artificial intelligence on copyrighted books without author permissions.
Getty initially brought suit against Stability AI, the creator of the prominent generative AI model Stable Diffusion, in January 2023. At the heart of Getty’s original argument was the contention that Stability AI had improperly trained their model using millions of copyrighted images without securing appropriate permissions.
Moreover, Getty alleged that many images generated by Stability’s AI closely resembled copyrighted originals, even depositing Getty’s distinctive watermarks on some outputs. However, these fundamental claims—centered both on Stability’s image-generation practices and the training process itself—were formally abandoned on Wednesday.
Legal experts indicated Getty likely faced difficulties proving jurisdictional connections to the UK copyright law in regard to training data, and in demonstrating substantial copying required for copyright infringement in AI-generated outcomes.
Getty’s team acknowledged weakening evidentiary support and inadequate expert witness testimony from Stability AI as key motivations for dropping their central allegations. They contend that streamlining their case allows stronger focus and a higher probability of success on remaining complaint elements.
One of the remaining arguments—secondary infringement—could prove critical, suggesting that AI models trained abroad may constitute infringing “articles” imported into the UK if their training involved unauthorized copyrighted content. This claim might significantly influence multinational AI developers.
In response to Getty’s decision, Stability AI expressed satisfaction, noting it remained confident that the remaining trademark infringement claims, particularly regarding watermarked images, would similarly fail under scrutiny. According to the company, users viewing watermarks on images created by Stable Diffusion do not perceive them as Stability AI’s commercial endorsement or brand identity.
Getty Images still faces a separate, sizable legal battle in the United States. Filed in February 2023, that lawsuit accuses Stability of unlicensed use of approximately 12 million copyrighted images and seeks statutory damages amounting to $1.7 billion.
In addition to these challenges, Stability AI remains entangled in other lawsuits, including one initiated by visual artists against Stability AI, Midjourney, and DeviantArt, alleging copyright infringement due to unauthorized use of artistic works in model training.
Getty itself has entered the generative AI scene with its proprietary tool—trained exclusively on licensed images from its own stock archives—which enables customers to create new, copyright-compliant visual content.
