US-UK On Fair Use

Why It Matters

The case sets a precedent for how courts may treat AI models and generated outputs as infringing works, influencing global enforcement strategies and prompting legislative action on AI‑driven data use.

Key Takeaways

• Getty Images sued Stability AI for training and output infringement.
• Primary copyright and database claims dropped due to jurisdictional evidence gaps.
• Remaining claims focus on model as infringing article and watermark misuse.
• UK Data Access Bill passed, but excludes explicit AI copyright provisions.
• Judgment pending; case highlights cross‑border AI training and enforcement challenges.

Summary

The webinar examined the rapidly evolving landscape of AI‑generated content and copyright law, centering on the landmark UK trial between Getty Images and Stability AI. The case, heard in the Patents Court—a specialist IP forum within the Chancery Division—originally featured six heads of claim, including primary copyright infringement for both training inputs and generated outputs, database rights, trademark infringement, and passing‑off allegations. During the 18‑day trial, Getty Images withdrew its primary copyright and database claims after evidence showed most model training occurred outside the UK and Stability AI altered its prompts to prevent reproducing disputed images. What remains are secondary claims that treat the Stable Diffusion model itself as an infringing article, alongside trademark and passing‑off claims tied to synthetic Getty watermarks that appeared in AI‑generated images. The presenters highlighted striking visual examples: AI‑produced pictures that replicated Getty’s watermark and even garbled credit lines, underscoring the technical difficulty of training on watermarked data. They also clarified the UK legislative backdrop, noting the Data Access and Use Bill’s passage—primarily aimed at open banking—while a separate government consultation debates a mandatory licensing regime to compensate artists for training data use. The unresolved judgment, expected later this year, will shape how courts address cross‑jurisdictional AI training, the status of AI models as copyrighted works, and the enforceability of trademark rights in generated content. Simultaneously, the UK’s pending policy discussions signal a broader move toward regulated AI data usage, with potential ripple effects for developers and rights holders worldwide.

Original Description

Generative AI is transforming creative industries—and igniting legal battles across the globe. As U.S. and U.K. high profile cases are testing the limits of fair use, courts are being asked to define how copyright, trademarks, and data rights apply in the age of machine learning. Join Rajvinder Jagdev (Powell Gilbert, London) and Dr. Chris Mammen (Womble Bond Dickinson, San Francisco) for a lively, cross-Atlantic conversation on the latest litigation, legislative developments, and what creators, tech companies, and lawyers need to know now.

► Homepage: https://www.law.berkeley.edu/research/bclt/

► View our CLE Catalog on our B-CLE Platform: https://bcle.law.berkeley.edu/

#BCLT #UCBerkeleyLaw