The Netherlands tests national AI as an alternative to US models

The Netherlands has begun practical testing of its national language model, GPT-NL. This project is a strategic step aimed at reducing Europe's dependence on American AI systems. After two years of development, the model is now being used in public services, municipalities, and forensic science. This is reported by Ixbt.com .

GPT-NL is designed not for ordinary users, but for the digitalization of public infrastructure. Currently, about 30 municipalities are increasing the efficiency of citizen requests for public services through a virtual assistant named Gem. The model also helps write official documents in simple, understandable language, facilitating communication between the state and citizens.

The most important aspect of the project is its approach to copyright. Unlike companies such as OpenAI or Google, the developers of GPT-NL have signed official license agreements with major publishers, newspapers, and TV channels in the country. This is the first collective agreement in the history of AI made with media outlets on a national scale.

Nevertheless, the project faces resource shortages. Compared to giants like OpenAI and Anthropic, a team of just 25 developers is working with a limited budget. Despite this, through this initiative, the Netherlands is demonstrating an alternative path for the development of transparent and secure AI under the control of state institutions.