The case of Getty Images (US) Inc. v Stability AI Ltd, now pending before the UK High Court, stands to set the nation’s first major precedent at the intersection of intellectual property rights and Artificial Intelligence model training. Getty Images alleges that Stability AI unlawfully scraped millions of its licenced photographs, using them and their metadata to train Stable Diffusion, a leading open-source generative model. Stability AI does not deny Getty content appears in its training data but claims the images were legally available, its technology produces only novel statistical outputs, and much of its development occurred on servers based outside the UK.
The dispute centers around several novel, complex legal issues. The primary question is whether training Artificial Intelligence on copyright-protected works without a licence constitutes ‘reproduction’ under the UK´s Copyright, Designs and Patents Act 1988. A decision in Getty’s favor could force developers to licence every protected work used in training sets, upending business models and Artificial Intelligence economics. The court must also clarify the territorial reach of UK copyright law, as Stability AI asserts that key activities were carried out abroad while Getty argues that commercial exploitation in the UK secures jurisdiction. Other noteworthy aspects include allegations of secondary infringement via downstream distribution, defences based on the CDPA’s pastiche exception for transformative works, and claims over the removal or stripping of digital metadata — each of which could shape how Artificial Intelligence models are built, released, and monitored going forward.
The litigation’s implications stretch far beyond the parties involved. A Getty victory would likely drive up training costs by requiring broad licensing or legal reform. It could prompt open-source Artificial Intelligence communities to adopt stricter controls on data provenance and risk management. On the compliance front, developers may soon need detailed copyright audits, robust data governance procedures, and rewired contractual protections throughout the Artificial Intelligence value chain. Meanwhile, the court’s decision on jurisdiction could make UK law reach into Artificial Intelligence activities carried out on foreign servers but commercialised in the UK. Businesses, developers, and rights holders are therefore closely watching the court’s handling of copyright, distribution, and data transparency as the outcome will impact not just the UK market but set precedents shaping global Artificial Intelligence innovation and regulation.
