Getty v Stability AI: a step forward or a stalled debate?

The case between a global leader in licensed stock photography and a UK based AI company, Getty Images (US) Inc. v Stability AI Ltd  was widely anticipated to provide the UK’s first major ruling on how copyright law applies to AI model training. However, in a significant shift during closing submissions, Getty withdrew its core claims - namely, those relating to primary copyright infringement and database rights. As a result, some of the most pressing legal questions remain unanswered.

While the case proceeds on narrower grounds, dropping the headline claims has reignited calls for legislative reform to clarify a legal area that’s evolving faster than the courts can keep up.

What changed in Getty Images (US) Inc. v Stability AI Ltd?

At the outset, Getty alleged that Stability AI unlawfully used millions of its copyright images to train its Stable Diffusion (open-source text-to-image) AI model. It also argued that the model’s outputs reproduced protected elements of its works - forming the basis of its primary copyright infringement claims.

These claims were withdrawn before judgment. Getty cited practical difficulties, including:

• Gaps in the evidence;
• A lack of witnesses with direct knowledge of the model’s training process; and
• An inability to link the alleged infringing activity to UK territory - something required under the Copyright, Designs and Patents Act 1988 (CDPA) to establish jurisdiction.

Getty also dropped its database rights claims, which were closely tied to the copyright claims. These required evidence of a substantial part of Getty’s image database being copied and that it qualified for protection under UK database rights law - again, something Getty wasn’t in a position to prove.

Which claims are still live in Getty Images (US) Inc. v Stability AI Ltd?

Three claims are continuing to trial in Getty Images (US) Inc. v Stability AI Ltd:

1. Secondary copyright infringement: Getty argues that distributing the Stable Diffusion model in the UK constitutes importing an “infringing article” under section 22 of the CDPA. This raises novel legal questions, including whether software models (like model weights) qualify as "articles" and whether they embody infringing material.
2. Trade mark infringement: Some outputs allegedly contained Getty’s watermarks or branding. The claim under the Trade Marks Act 1994 focuses on whether such marks were used in the course of trade, without permission, and in a way that could confuse consumers.
3. Passing off: Getty contends that outputs showing its branding could mislead the public into thinking there's an association or endorsement, which could damage its reputation and goodwill.

Why did Getty withdraw its key claims in Getty Images (US) Inc. v Stability AI Ltd?

1. Territorial scope: Stability AI maintained that all training occurred outside the UK (primarily in the US). Getty wasn’t able to show that any copying took place within the UK jurisdiction, and its attempt to amend its pleadings to reflect new evidence was too late.
2. Evidential limitations: Getty faced difficulties tracing a direct line from its content to the training data or outputs. In AI systems, outputs often blend influences in non-obvious ways, making it hard to show that a “substantial part” of a protected work has been copied - an essential test under UK copyright law.
3. Challenging output claims: Getty relied on examples where generated images resembled its content or included its watermark. However, demonstrating that this amounted to infringement was challenging, particularly where there was no direct copying but rather algorithmically generated similarities.
4. Prompt engineering concerns: Stability argued that Getty may have used tailored prompts designed to produce infringing outputs, which could weaken the case - especially if the results wouldn’t ordinarily appear during typical use.
5. Reduced risk of further harm: Stability has since implemented prompt filters to block certain terms, including Getty’s name. This was likely taken into account in assessing whether an injunction was necessary or proportionate.

What does the reduced claims in Getty Images (US) Inc. v Stability AI Ltd mean for traditional IP rights and generative AI?

This case was expected to offer guidance on whether training AI models on copyright content without a licence breaches UK copyright law. With those claims withdrawn, we are no closer to an answer.

Instead, the focus now shifts to the secondary infringement claim - and whether AI models themselves, or their distribution, can be considered infringing “articles” under UK law. This is untested legal territory and could have ripple effects beyond this case.

The trade mark and passing off claims may also give rights holders some direction on how branding and consumer perception intersect with AI-generated outputs.

What this means for rights holders, AI developers, users and deployers?

While we await the court’s decision, there are new key takeaways for different groups potentially impacted to consider:

• For developers: It's crucial to keep records of where and how training occurs. Ensuring that no UK-based activity crosses into infringement territory is an important part of managing legal risk.
• For rights holders: Ownership, originality, and use all remain central. Where possible, metadata and visible branding can help track and assert rights over content used in training or output.
• For AI users and deployers: Be aware of potential secondary liability. Review your contracts for IP warranties and indemnities, particularly if the outputs are client-facing or commercial.

Next steps

A decision is expected later in 2025. The judge may comment on the withdrawn claims, but these remarks will not set a legal precedent, but the remaining claims may still produce valuable guidance.

Under the Data (Use and Access) Act 2025, the UK government is required to publish a full economic assessment and policy report on the intersection of AI and copyright by March 2026. While pressure for legislative clarity predates this case, Getty’s withdrawal of key claims has added further weight to calls for reform.

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.