Home / News & insights / Insights / Will Getty v. Stability AI produce UK’s first major precedent in IP law and AI model training?

Will Getty v. Stability AI produce UK’s first major precedent in IP law and AI model training?

Posted on in Commercial and Tech

Will Getty v. Stability AI produce UK’s first major precedent in IP law and AI model training?

Generative AI is reshaping innovation across industries, resulting in an unsettled legal framework governing how these systems are trained. However, one case may soon begin to define the boundaries: Getty Images (US) Inc. v Stability AI Ltd, currently before the High Court in London.

It raises pivotal questions around copyright, jurisdiction, and data provenance - and could become the UK’s first major precedent on how intellectual property law applies to AI model training.

The outcome will likely shape how AI developers, rights holders, and businesses structure their models and datasets, not just in the UK but across global markets.

The case: Getty Images (US) Inc. v Stability AI Ltd

Getty Images, a global leader in licensed stock photography, alleges that Stability AI, a UK based AI company, unlawfully scraped millions of its images and associated metadata to train Stable Diffusion, its open-source text-to-image AI model, in breach of Getty’s intellectual property rights.

Stability AI accepts that some Getty images appear in its training data but denies liability. It argues that:

  • Images were lawfully accessible online;
  • Stable Diffusion generates statistically novel outputs rather than reproducing works; and
  • Many training activities occurred on servers outside the UK.

Key legal issues raised in Getty Images (US) Inc. v Stability AI Ltd

This case raises several novel and complex legal issues:

  • Does training an AI model constitute “reproduction”?

The Court will have to decide whether feeding copyright works into a machine-learning pipeline, without a licence, constitutes infringement under the Copyright, Designs and Patents Act 1988 (CDPA).

A ruling in Getty’s favour would require developers to obtain licences for all protected works in a training set - fundamentally altering both the approach and the economics of AI development.

  • Jurisdiction – when does UK copyright law apply?

Stability AI claims that development and model training took place on non-UK servers. Getty argues that commercial exploitation and distribution occur in the UK, giving English court jurisdiction. The court’s handling of territorial reach will clarify how far English law extends to AI activity hosted abroad.

  • Secondary infringement

Getty argues that making Stable Diffusion available in the UK via GitHub, HuggingFace and DreamStudio imported an infringing “article”. The judgment should clarify how far liability travels down the AI value chain. A consideration of vital interest to open-source ecosystems.

  • Pastiche defence [s.30A of Copyright, Designs and Patents Act 1988 (CDPA]

In the alternative, Stability AI relies on the pastiche exception, contending that its outputs are transformative, non-substitutive and unlikely to compete with the originals. The case seems likely to provide the first UK judicial guidance on applying pastiche to generative AI.

  • Removal of metadata

Getty also alleges that Stability AI stripped or failed to preserve metadata identifying ownership and licensing terms, contrary to the CDPA. This is an independent ground of liability and reflects the growing importance of data provenance and transparency in AI systems.

Why this case matters

This litigation may fundamentally redefine how copyright is interpreted and enforced in the AI age. Key implications include:

  • Licensing costs for AI training: a ruling in Getty’s favour could impose significant licensing expenditure - or compel statutory reform - to clear millions of images for model-training purposes.
  • Open-source chilling effect: Stability AI’s model is open-source. If found liable, it may deter similar initiatives or prompt stricter controls on data provenance.
  • Compliance and data governance: developers may need end-to-end lineage tooling and contractual indemnities throughout the AI value chain.
  • Jurisdictional reach: the court’s jurisdictional analysis may extend UK enforcement to overseas model-training where UK exploitation follows - an issue of growing significance in a globalised tech landscape.

Practical takeaways for AI developers, rights holders and distributors

AI developers, rights holders, and distributors should closely monitor the outcome of this case. There are a range of steps to consider exploring while this case progresses:

  • Conducting copyright audits of datasets used in training.
  • Reviewing contracts for licensing and indemnity protections.
  • Reevaluating risk exposure from using third-party content in AI training.
  • Staying abreast of emerging case law and regulatory reforms.

For rights holders, the case also reinforces the need to actively monitor digital content, assert legal rights where appropriate, and ensure content is clearly watermarked or tagged with enforceable metadata.

Conclusion

Getty Images v Stability AI is more than a commercial dispute - it is a potential landmark test for how traditional IP rights will intersect with new technologies, particularly generative AI. The High Court’s ruling will likely influence how businesses across the AI spectrum structure their data strategy, distribution, compliance and risk management models.

Hamlins Commercial and Tech team support clients across multiple industries, sharing commercial and regulatory expertise and advising on the protection of intellectual property rights and AI use. If you would like a conversation about how we can help you navigate changes impacting your business, please get in touch.