This month, courts in the United Kingdom and Germany issued key decisions regarding the use of protected works to train artificial intelligence models. In the United Kingdom, Getty Images sued Stability AI for using millions of images without permission and for including watermarks in AI-generated images. Although the Supreme Court stated that there was insufficient evidence of direct reproduction of the works, thus rejecting the copyright infringement claim it confirmed trademark infringement when Getty’s watermarks appeared in the generated images.
This British ruling set an important precedent regarding the burden of proof in AI-training cases, making it clear that merely showing access to protected works is not enough; identifiable reproduction must be demonstrated. However, the use of protected distinctive signs, such as watermarks, does create legal liability due to confusion or improper association, even when there is no literal copy of the original image.
In Germany, the Society for Copyright Management sued ChatGPT for reproducing fragments of protected song lyrics. The Munich Court concluded that the AI stored portions of these works and reproduced them verbatim when given certain prompts, constituting direct infringement. As a result, the court ordered an end to this practice and the payment of damages, reaffirming that musical works are also protected against unauthorized use by AI tools.
