Dutch Publishers Support Ethical Training of AI Model, Partnering for GPT-NL

In a innovative move poised to reshape the future of artificial intelligence, Dutch news publishers have joined forces with the Netherlands Organisation for Applied Scientific Research (TNO) to develop GPT-NL, a large language model meticulously trained on ethically sourced, legally obtained Dutch data. This collaborative initiative, the first of its kind globally, sees over 30 national and regional publishers, including members of NDP Nieuwsmedia and news agency ANP, actively contributing their archived articles to ensure the AI’s development aligns with crucial ethical standards.

The partnership aims to double the volume of high-quality Dutch training material available for GPT-NL, fostering a robust and diverse foundation for the model. Crucially, the project prioritizes transparency, copyright adherence, and fair compensation for content contributors. This stands in stark contrast to practices by some international AI developers who have trained models on vast amounts of internet-scraped content without permission or remuneration.

Selmar Smit, Manager of Science & Technology at TNO and a founder of GPT-NL, expressed pride in the collaboration, emphasizing that it sends a strong message: “AI can be developed responsibly, with respect for copyright and public values.” Rien van Beemen, Chair of NDP Nieuwsmedia, echoed this sentiment, highlighting that the partnership sets a precedent for ethical AI innovation that strengthens journalism in the Netherlands and avoids the unauthorized use of journalists’ work.

GPT-NL is a non-profit endeavor, developed in collaboration with the Netherlands Forensic Institute (NFI) and SURF, and is envisioned as a responsible alternative to existing large international AI systems. The model, currently in training since June 2025, is designed for specific tasks such as summarization, simplification, and information extraction, and strictly adheres to European legal frameworks, including the forthcoming AI Act.

Strict safeguards are being implemented to prevent unauthorized extraction or reuse of content once the model is released in the fourth quarter of 2025. This commitment to data protection, digital sovereignty, and ethical considerations in AI development positions the Netherlands at the forefront of responsible technological advancement, ensuring that AI innovation serves societal good while upholding fundamental rights and values.