Statement on commercial generative AI

As a national library, the KB bears great responsibility for the work of writers and publishers which we make accessible via channels such as our online platforms. The KB embraces digital developments and is excited about the opportunities that Artificial Intelligence (AI) can bring to society but at the same time, we believe it is very important for AI companies to operate in an ethically responsible manner.

The KB notes that commercial parties who crawl digital resources on websites on a large scale for training models, using applications such as ChatGPT, are not complying with the AI principles established by the KB in 2020. 

The KB makes written resources widely accessible and expects re-users to respect copyright, indicate sources and process personal data responsibly. In our opinion, AI companies are failing in that respect. In our view, the way many commercial AI models are now trained – by crawling websites without permission – does not serve the public interest. Thus we want to prevent the rights that writers and publishers have over their publications from being harmed by this type of re-use. We have therefore adapted our terms of use and taken technical measures to combat crawling of our websites Delpher and DBNL by commercial parties from now on. 

It remains our policy to provide access to information as openly as possible. An important goal of the KB is to encourage academic research based on our collections as much as possible. We guarantee that this reuse shall not be hindered by our measures against AI companies. Moreover, we will closely monitor developments in AI, as well as case law and further European and national regulations. 

On this basis, we will therefore continue to monitor and evaluate our AI policy and make adjustments where necessary. Not only because we want to be transparent in the way we deal with AI ourselves but also to show how others can handle the resources that we make accessible via our websites in a responsible manner.