With a new guide, the Rights Alliance aims to equip Danish companies and institutions to develop artificial intelligence with respect for copyrights
Creative content such as literature, movies, series, music and articles are valuable training data among developers of generative artificial intelligence. However, the Rights Alliance’s experience shows that the use of protected content for the development of artificial intelligence is often illegal.
Therefore, the Rights Alliance is publishing a new guide to guide Danish institutions and companies on how copyrighted content can be used legally to train artificial intelligence.
Danish companies need to be educated as well
Public coverage of the illegal use of creative content to train artificial intelligence has so far focused on The New York Times’ case against OpenAI and the Rights Alliance’s takedown of the illegal Books3 dataset, which several tech companies, including Meta, Bloomberg and StabilityAI, have used to train their AI models. But the cases don’t stop at US borders.
In Denmark, the Rights Alliance has also seen cases of AI being trained with illegal content.
Recently, it was announced that the Alexandra Institute, IBM Denmark, and the Danish Chamber of Commerce have joined forces to form the Danish Language Model Consortium, where Danish companies commit to developing artificial intelligence responsibly. The Rights Alliance is therefore making the guide available to the Danish Language Model Consortium – and AI developers in general – and hopes that it will raise awareness of the respect for rights when artificial intelligence is developed based on content created by the creative industry.
Read more: Rights Alliance removes the illegal Books3 dataset used to train artificial intelligence