Part 10: Copyright And AI: Responsibility Of Providers And Users

Most people who regularly work with digital devices have already used AI applications to generate output - for example, a title for a blog post. In order for the AI application to be able to do this, the underlying model must be trained in advance with large amounts of data which often include copyrighted content.  In this post no. 10 of our AI blog series, we examine the question of whether copyright-protected works may be used to train AI models, what the consequences are if a copyright-protected work is still recognizable in the output generated by an AI application and who - the provider or the user - is responsible.

The current debate on AI and copyright

AI applications generate a lot of new content, but also require an enormous amount of content input to do so. From a copyright perspective, the following key questions arise: