RSF hails media copyright requirements in European Parliament’s AI Act
Reporters Without Borders (RSF) welcomes the version of the proposed Artificial Intelligence Act that has been adopted by the European Parliament, as it would require language model producers such as OpenAI to publish summaries of the copyrighted content used to develop their products, which would benefit the media. Producers of large language models should make this information public now, RSF says.
After adopting its negotiating position on the AI Act on 14 June by 499 votes in favour, 28 against and 93 abstentions, the European Parliament will now negotiate the law’s final form with the Council of the European Union and the European Commission.
The version approved by the parliament would place transparency requirements on AI designers and producers such as OpenAI. Before releasing large language models such as GPT-4, the model used to produce the ChatGPT chatbot, they would have to publish a detailed summary of the use of training data protected under copyright law.
“We are delighted by the determined stance taken by the European Parliament and we hope that the EU’s other bodies will move in the same direction. We call on producers of large language models, such as OpenAI, to make the copyrighted material used to design their technologies public now. Publicly listing the journalistic content subject to copyright that is used to produce AI is a necessary condition for laying the foundations for a relationship of trust between the media and AI producers.”
Vincent Berthier
Head of RSF’s Tech Desk
Journalistic content is of great value to language model producers because of its editorial quality and the reliability of its information. It nourishes these technologies and improves their performance. However, the media’s viability is directly threatened by some of the uses to which they are put, such as AI-generated content farms or chatbots integrated into search engines that plunder news sites for their content. By collecting journalistic content in an unregulated manners, sites created by AI can also capture much of the revenue that should have gone to media outlets.
Tech companies must show that they are able to understand these issues, propose agreements with media publishers and help develop an ecosystem that benefits the media.
