UK Court Finds Stability AI Free of Copyright Liability but Citing Trademark Infringement
Key Points
- UK judge finds Stability AI did not infringe Getty's copyright.
- Court rules the AI model violated Getty's trademark rights.
- Stability AI does not store or reproduce protected works.
- Responsibility for trademarked content lies with the model provider.
- Getty declares the trademark finding a victory for IP owners.
- Stability AI emphasizes that most copyright claims were dismissed.
- The ruling is narrow and specific to this case's evidence.
- Similar AI copyright cases in the US have produced mixed results.
In a London courtroom, Justice Joanna Smith concluded that Stability AI, the maker of Stable Diffusion, did not violate copyright law when training its image models on Getty Images' content. The ruling emphasized that the AI system does not store or reproduce protected works. However, the judgment noted that Stability AI breached Getty's trademark rights by allowing users to generate images resembling Getty and iStock logos. The decision is narrowly scoped, offering a mixed outcome for both the AI developer and intellectual‑property owners.
Background of the Case
A lawsuit brought by Getty Images against Stability AI centered on claims that the AI company scraped millions of Getty‑owned photographs to train its popular Stable Diffusion model. Getty argued that this practice infringed both copyright and trademark protections.
Justice Joanna Smith's Ruling on Copyright
The UK court determined that Stability AI did not infringe Getty's copyright. Justice Smith explained that the AI system does not "store or reproduce any Copyright Works" and therefore does not meet the legal definition of infringement. This portion of the decision was described as both historic and extremely limited in scope, reflecting the nuanced approach courts are taking toward AI‑related disputes.
Trademark Findings
While dismissing the primary copyright claims, the court found that Stability AI violated Getty's trademark rights. The ruling noted that the AI model allowed users to generate images that included elements resembling the iStock and Getty Images logos. According to the judgment, responsibility for the presence of such trademarks lies with the model provider, which controls the training data.
Implications for Stability AI and Users
The decision means that Stability AI can continue operating in the United Kingdom without the threat of copyright liability. However, the company must address the trademark issue, as the court rejected its attempt to shift responsibility onto individual users. Stability AI’s counsel highlighted that the ruling resolves the core copyright concerns originally raised.
Reactions from the Parties
Getty Images hailed the outcome as a win for intellectual‑property owners, emphasizing the trademark infringement finding. Stability AI pointed out that the dismissal of most copyright claims left only a subset of issues before the court, and that the final ruling resolves the core copyright dispute.
Broader Context
This case is one of the first major AI lawsuits involving a large content library. Similar disputes in the United States have seen mixed results, with companies like Anthropic and Meta prevailing over authors alleging unauthorized use of their works. The UK decision adds to a growing body of case law that is shaping how courts interpret existing copyright and trademark statutes in the age of generative AI.
Future Outlook
Justice Smith stressed that the ruling is specific to the evidence and arguments presented in this case, leaving open the possibility of different outcomes in future lawsuits that involve other statutes or factual contexts. The decision underscores the need for AI developers to carefully manage trademarked material in their training datasets while navigating an evolving legal landscape.