How TNO is leading the drive towards sovereign, responsible Dutch AI

Addressing an urgent need

According to Berry Vetjens, Director of market unit ICT, Strategy & Policy (ISP) at TNO, the time for European organisations to establish trustworthy AI practices is right now. “AI is here to stay, and it is not an option to not participate in its responsible development,” he says. “Otherwise, the market will organise itself into a system that does not suit our European values.”

Among those European values are privacy, transparency, sustainability, and security – elements currently lacking in the most popular non-European AI models. That’s why TNO is dedicated to a vision of Responsible AI that Works: sovereign, independent, responsible, and trusted AI models that are not only effective, but also adhere to European values. To achieve that vision, TNO is currently active in three distinct pathways. “We develop, we share, and we deploy responsible AI expertise to accelerate this AI vision in the Netherlands and beyond.”

1. Developing responsible AI

For years, TNO has been developing both the technology and know-how to develop effective AI models. Whether it’s narrow AI developments that perform specific functions, or general models that learn and develop over time, understanding the principles of AI development is key to their responsible use.

With the upcoming release of GPT-NL – the Netherlands’ first Large Language Model – TNO aims to prove that effective models can be deployed at scale. GPT-NL is sovereign to the Netherlands, legally compliant with EU regulations for privacy and security, and adheres to strict guidelines for responsible data usage. “With GPT-NL, we show that TNO has the intelligence needed to build these models, and the wisdom needed to build them right,” Berry says.

2. Sharing critical knowledge

That intelligence and wisdom is only useful if it is shared with those who need it most. Which is why the second pathway to TNO’s AI vision is about knowledge transfer. “We want as many organisations as possible to benefit from our knowledge and expertise,” Berry explains. “The key to successful AI development is not just computing power – it’s expertise in responsible model development. So, we aim to accelerate the adoption of responsible AI principles.”

One of the ways TNO will achieve this is by acting as the dedicated Expertise Centre for the Netherlands’ first AI Factory: a knowledge centre and supercomputing facility where organisations can build and test their AI models. “Our contribution to the AI Factory is especially relevant for SMEs, who normally may not have access to TNO’s expertise. At the factory, we’ll help them develop the skills and knowledge to deploy their own responsible AI models.”

One of the ways TNO will achieve this is by acting as the dedicated Expertise Centre for the Netherlands’ first AI Factory: a knowledge centre and supercomputing facility where organisations can build and test their AI models. “Our contribution to the AI Factory is especially relevant for SMEs, who normally may not have access to TNO’s expertise. At the factory, we’ll help them develop the skills and knowledge to deploy their own responsible AI models.”

3. Deployment with a difference

AI systems are not used in isolation. They support and supplement existing technologies and interactions, extract data from protected sources, and influence the decision-making of the people who use them. So, effective deployment is the key to maintaining a safe and secure AI ecosystem. By supporting effective AI deployment, TNO will prepare Dutch organisations to extract the full value of AI in a business environment. “We aim to ensure that every AI deployment is trustworthy and secure. That organisations know what can happen when they connect AI into their systems, and how to do it the right way to get the intended effect and avoid the potential dangers,” Berry says.

Looking even further into the future, Berry envisions a time when federated AI systems connect across Europe to further multiply their power and potential. “The more responsible, secure, sovereign European models we can join up, the more Europe can benefit from them. “We are already supporting responsible deployment through our AI Oversight Lab, where we assess models for bias, transparency, and compliance. We are also ready to extend our technological expertise to enable organisations to deploy effectively and responsibly across the Netherlands and Europe.”

Smart technology, wise implementation

The age of AI is upon us. And TNO aims to ensure that all the values that define Europe are safeguarded. Amidst the growing energy for AI’s potential, TNO is leading the drive towards incorporating the right values into every AI development. “We must give attention to all three pathways: develop, share, and deploy,” Berry says. “Otherwise, the impact will be diluted. For TNO, it is not a maybe but a must: the Netherlands will have sovereign and responsible AI models both now and in the future, and TNO will drive that. Organisations can trust us to help them do it, and do it right.”