Dutch AI project GPT-NL enters real-world testing (www.techradar.com)

The Netherlands has initiated real-world testing of its national AI model, GPT-NL, marking a significant step toward reducing reliance on Silicon Valley AI systems. Developed over the past two years through collaboration between government agencies and research organizations, GPT-NL aims to serve practical public sector applications, such as government communication and municipal assistance. This move is not only about creating a competitive alternative to existing technologies but also about ensuring European compliance with legal standards and securing local control over data infrastructure. One of the project's standout features is its licensing agreements with Dutch news publishers, making GPT-NL the first AI initiative globally to secure collective agreements in a single market. This legal framework addresses the contentious issue of copyrighted training data, fostering a more amicable relationship between AI systems and journalism. Initial pilots include a virtual assistant to enhance citizen interactions and a writing aid for civil servants, highlighting the model's focus on practical utility over flashy consumer applications. However, the project faces challenges regarding funding and resources, as it operates with a modest budget and a small team, raising questions about its long-term viability in keeping pace with global AI advances.