Getty v. Stability (UK)

The claims

Copyright infringement (primary claims later withdrawn), secondary infringement by importing an "infringing copy", database right, trademark infringement, passing off

What has happened

Getty sued Stability AI in London in January 2023, alleging Stable Diffusion was trained on millions of Getty photographs scraped without a licence and that outputs reproduced its works and watermarks. The case shrank sharply at trial. In June 2025 Getty withdrew its primary training and output claims after the evidence showed the training took place outside the UK, putting those acts beyond the territorial reach of UK copyright law. In her November 4, 2025 judgment, Mrs Justice Joanna Smith dismissed the remaining secondary-infringement theory, holding that a model which never stored or reproduced the underlying works is not an "infringing copy" under section 27 of the Copyright, Designs and Patents Act. Getty won only narrow, largely historic trademark findings where Getty and iStock watermarks appeared on images generated by early Stable Diffusion versions. In December 2025 the judge granted Getty permission to appeal the secondary-infringement ruling in full, with an appeal to be filed by February 3, 2026.

Key developments

• 2023-01 — Getty issued High Court proceedings over the training of Stable Diffusion and its outputs, alongside its parallel US case.
• 2025-06-09 — Trial opened before Mrs Justice Joanna Smith in the Chancery Division.
• 2025-06-25 — Getty withdrew its primary copyright claims over training and outputs mid-trial, after evidence indicated training occurred outside the UK.
• 2025-11-04 — Judgment handed down: secondary-infringement claim dismissed — an AI model is not an "infringing copy" under s.27 CDPA — with only narrow trademark findings on generated watermarks in early model versions.
• 2025-12-16 — At the consequentials hearing, the court granted Getty permission to appeal the secondary-infringement dismissal in full, with the appeal to be lodged by February 3, 2026.

Why it matters for training data

The first full trial of an AI training dispute in a major jurisdiction ended without a ruling on the core question, because the training happened outside the UK. Where a model is trained now determines which country’s law can reach it, and buyers of models trained abroad cannot assume UK law is settled: if the appeal succeeds, importing model weights trained on unlicensed data could itself infringe. The trademark findings also show that watermark remnants in outputs create liability separate from copyright, which makes provenance and watermark hygiene a supplier-side obligation.

Sources

• Judiciary of England and Wales — judgment page
• Latham & Watkins analysis of the High Court ruling
• Herbert Smith Freehills Kramer on permission to appeal
• Taylor Wessing on next steps for the appeal
• CMS on the consequentials hearing

Deeper analysis

• Getty v. Stability, explained

← All case files Weekly lawsuit news