Getty and Stability AI face off in landmark UK copyright trial

For the past two years, AI labs have operated on a simple playbook: scrape first, claim fair use later. That playbook just met its first real judge.

On Monday, Getty Images and Stability AI faced off in London's High Court for the generative AI industry's first major copyright trial. The trial is scheduled to run for three weeks, with a written ruling to follow at a later date.

Getty initially sued the London-based startup in both the US and the UK back in early 2023. While American lawsuits against AI companies are still tangled in procedural motions, the UK court moved ahead. Getty claims Stability carried out brazen infringement on a staggering scale by scraping millions of photos from its catalog to train Stable Diffusion.

A battle over fair use and free data

Getty came out swinging on day one. Trial lawyer Lindsay Lane told the court that Stability had a voracious appetite for images and was completely indifferent to whether they carried watermarks, copyrights, or NSFW content. Lane argued that Stability rushed to beat rivals to market, calling the trial a day of reckoning for that cut-corners approach.

Getty is trying hard to avoid looking anti-tech. Lane argued the lawsuit is not a fight between creatives and technology that would end AI, but a straightforward fight over paying for ingredients. Getty CEO Craig Peters previously insisted that creators must be asked for permission up front, rather than forced into an opt-out regime.

Stability, meanwhile, is framing the case as an existential fight. In legal filings, lawyer Hugo Cuddigan argued that Getty's claims represent an overt threat to Stability's whole business and the broader generative AI sector. A spokesperson added that artists using AI build upon collective human knowledge, which sits at the core of freedom of expression.

Stability is also leaning on technical technicalities. It argued that the training technically took place on Amazon servers located outside the UK, meaning British courts lack jurisdiction. It also insisted that only a tiny fraction of its model's actual outputs look anywhere close to Getty's original photos.

Why it matters

This trial is the first real legal stress test for the generative AI boom, and the outcome will ripple far beyond London. For creators and stock agencies, a Getty win would end the era of involuntary donation, effectively forcing AI developers to negotiate paid licensing deals rather than hiding behind opt-out forms.

For AI startups, the free lunch looks like it is officially over. Stability has pulled in high-profile backing, including Facebook veteran Sean Parker as board chair and an investment from ad giant WPP, but defending foundational training methods in court is an expensive game. If courts reject the idea that web scraping automatically qualifies as fair dealing, the cost of training frontier models will skyrocket, leaving only deep-pocketed tech giants capable of paying upfront content fees.

We think Getty's legal positioning is particularly sharp: by framing copyright licensing as essential fuel for AI rather than an obstacle, it forces Stability to defend raw scraping on purely economic grounds. Stability's jurisdictional defense about cloud servers feels more like a delay tactic than a sustainable answer. As outside legal observers have pointed out, the court is unlikely to grant blanket copyright exemptions to AI firms, meaning rights holders are about to gain massive leverage at the bargaining table.