Getty vs. Stability AI: A Landmark Lawsuit on Copyright and Artificial Intelligence.

Gianpaolo Todisco

On June 9, 2025, the highly significant legal case between Getty Images and Stability AI began before the High Court in London. This is one of the first legal battles in the world regarding the use of copyrighted material. copyright to "train" generative AI models (like Stable Diffusion).

Getty's accusations

Getty claims that Stability AI downloaded millions of images from its unlicensed archive and used them to train Stable Diffusion. The main allegations include:

  • Violation of the copyright , for unauthorized use of protected works.

  • Trademark infringement, as the AI ​​would have generated images with the Getty watermark.

  • Database copyright infringement, as Getty also claims rights to its vast photographic archive

Stability AI's defense

Stability AI, based in London, denies the accusations. It argues that the training took place on servers in the United States and is therefore not subject to British jurisdiction. It also argues that the system fosters creative innovation and freedom of expression and that very few generated images actually resemble the protected originals.

Legal and cultural impact

The trial will last several weeks, with a final decision expected later in 2025. Lawyers agree that this ruling could:

  • Defining the legal boundaries for the use of protected content in AI.

  • Influence negotiations for commercial licensing of material to creatives.

  • Determine the attractiveness of the UK as a hub for AI development, based on the stringency of safeguards copyright .

Evolution of the litigation

Internationally, Getty had already filed a similar lawsuit in the United States in February 2023, seeking damages of up to $1.7 billion for over 11,000 images. Recently, however, Getty dropped some of its claims of direct infringement. copyright in the UK, focusing on secondary and trademark rights.

The legal context is enriched: in parallel, in California, a judge has established that the training of Anthropic models on potentially pirated books does not constitute infringement of copyright All of this complicates the prospects of creatives seeking legal protection for their work.

Back
Back

Deepfakes and Digital Identity: Denmark's Revolutionary Proposal on Copyright.

Next
Next

Blanding: Between Minimalism and Legal Risks