Getty Images v. Stability AI: Undisclosed Litigation

The first major western judgment on generative AI and IP: Getty’s UK claims over Stable Diffusion’s training on millions of its images largely failed at trial, with only a narrow trademark win on watermark outputs.

LATEST DEVELOPMENT · APR 23, 2026

At the December 2025 consequentials hearing, Getty was granted permission to appeal the ‘infringing copy’ ruling to the Court of Appeal, while Stability was refused permission to challenge the narrow trademark findings. In the US, Getty’s refiled Northern District of California case largely survived Stability’s motion to dismiss on April 23, 2026, only a DMCA false-CMI claim was dismissed, with leave to amend, and is proceeding, with no trial date yet set.

PREDICTED OUTCOME

AI-GENERATED · METHODOLOGY →

Settlement ~45% likely HORIZON: global resolution most likely between late 2027 and… CONFIDENCE: MEDIUM

The most likely endgame is a negotiated global resolution, roughly 45 percent, landing between late 2027 and mid-2028 after the Court of Appeal rules in London and US discovery in N.D. Cal. matures. Expect a package rather than a pure damages payment: a retrospective release, a forward content licence for Getty’s library, watermark and output filtering commitments, and cash or equity likely in the $25 million to $150 million band, well short of the statutory-damages headline Getty could theorise. The driver is asymmetric endurance. Stability is thinly capitalised and cannot fund parallel transatlantic litigation into a 2028 jury trial, while Getty, having lost the training claims in the UK on situs grounds and lost secondary infringement on the merits, now has a licensing-market strategy that a deal serves better than another adverse precedent. The Anthropic authors settlement showed that AI defendants pay to close copyright exposure once dismissal fails, and Getty’s US case survived almost intact on 23 April 2026, which restores just enough leverage to price a deal. The leading alternative, about 25 percent, is a clean respondent outcome in which the Court of Appeal affirms that model weights holding no copies are not infringing copies and the US claims are then narrowed to a token trademark verdict. A Getty win on the UK appeal is the tail that would reprice everything upward, but the statutory text of the CDPA makes that the less likely reading at roughly 30 percent, and even success there only returns the case to a fight over remedies.

Generated by claude-opus-5 on Aug 4, 2026, from this case’s record only. It stands until a major development changes the record. Not legal or investment advice. How predictions are made →

BACKGROUND

Getty alleged that Stability AI scraped over 12 million Getty photographs to train Stable Diffusion, infringing copyright, database rights and trademarks, the flagship test of whether training generative models on scraped content is unlawful. Because the training occurred outside the UK, Getty dropped its primary training-infringement claims mid-trial and pressed secondary infringement and trademark theories.

In November 2025, the High Court rejected the secondary copyright infringement claim (holding that an AI model which stores no copy of the works is not an ‘infringing copy’) and gave Getty only a limited win on trademark infringement where outputs reproduced Getty watermarks. Getty has permission to appeal that copyright ruling, and the parallel US litigation, voluntarily refiled in the Northern District of California in August 2025, does reach the training conduct and continues, as do dozens of other AI training cases worldwide.

TIMELINE

APR 23, 2026PROCEDURAL

N.D. Cal. largely denies Stability’s motion to dismiss

Trademark infringement, false designation, dilution and unfair competition claims proceed, and Getty’s copyright claims remain in the case. Only the DMCA false-CMI claim was dismissed, with leave to amend.
DEC 16, 2025APPEAL

Getty granted permission to appeal to the Court of Appeal

Permission granted in full on the secondary infringement (‘infringing copy’) point as novel and important. Stability was refused permission to appeal the trademark findings.
NOV 4, 2025JUDGMENT

UK judgment: secondary copyright claim fails. Narrow trademark win

[2025] EWHC 2863 (Ch): model weights that store no copies are not ‘infringing copies’ under the CDPA. Extremely limited trademark findings on watermark outputs. The s.10(3) claim dismissed.
AUG 14, 2025PROCEDURAL

US case voluntarily dismissed in Delaware and refiled in the Northern District of California

No. 3:25-cv-06891, assigned to Judge Trina L. Thompson. The refiled complaint keeps copyright, DMCA and trademark theories and adds a market-dilution theory of harm.
JUN 25, 2025PROCEDURAL

Getty drops primary training-infringement and database-right claims mid-trial

Getty accepted it could not show the training of Stable Diffusion occurred in the UK, leaving secondary infringement and trademark claims for judgment.
JUN 9, 2025HEARING

UK trial opens before Mrs Justice Joanna Smith
FEB 3, 2023FILING

Getty files parallel US suit in the District of Delaware

Getty Images (US), Inc. v. Stability AI, Inc., No. 1:23-cv-00135 (D. Del.).
JAN 17, 2023FILING

Getty announces High Court proceedings against Stability AI in London