GPT-NL: the AI language model of the Netherlands

The Netherlands is developing its own AI language model: GPT-NL. This marks a significant step towards transparent, fair, and auditable use of AI. Its development will be funded by the Ministry of Economic Affairs and Climate Policy (EZK). But what can we expect?

Artificial Intelligence (AI)
March 7, 2024
Rik van den Buijs

What is GPT-NL exactly?

A collaboration of non-profit organizations, including TNO, SURF, and the Netherlands Forensic Institute (NFI), has announced plans for developing an open Dutch language model named 'GPT-NL'. This model will be accessible to academic institutions as well as government agencies, including those active in health, defense, and security sectors.

The goal of GPT-NL is to strengthen the digital autonomy of the Netherlands by deploying transparent, fair, and auditable AI. Selmar Smit, chief scientist at TNO, emphasizes that GPT-NL will not only contribute substantively to Dutch organizations but also support decision-making processes in accordance with Dutch and European values, leveraging technology and data.

What is the investment?

Approximately €13.5 million is allocated for the model's development. When comparing budgets, American tech companies are not considered direct competitors, given that OpenAI's value is estimated at $90 billion.


GPT-NL developer Selmar Smit indicates a reluctance to compete with billion-dollar AI models. Smit explains their intent to develop a much fairer and more responsible model. Transparency is crucial, and individuals may object if they disagree with certain data being used.

How will privacy be ensured?

GPT-NL will be trained using publicly available data, which may include personal information. To safeguard citizen privacy, the consortium has agreed to exclude non-anonymized data from the training process. While a 100% exclusion cannot be guaranteed, the consortium is exploring solutions such as an 'opt-out' option to assure citizens that their data will not be included in GPT-NL datasets.

The Netherlands Forensic Institute (NFI) underscores the importance of transparency in using models for forensic research. The consortium aims to provide clear explanations of the model's operation and training to all parties involved in the criminal justice process. These efforts aim to minimize the risk of subjectivity and unconscious biases in NFI investigations.

The first version of GPT-NL is expected to be ready by the end of this year.

This site uses anonymous cookies. Click on "Agree" if you agree to the use of cookies, or click on "Change" to determine your preferences.
This site uses anonymous cookies.