Lessons from the Ministry of Justice and Security on the responsible use of AI
Many senior leaders still view AI primarily as a technological development. According to Michel van Leeuwen, Director for AI, Data Protection and Privacy at the Dutch Ministry of Justice and Security (JenV), this is a misconception. “AI doesn’t just affect the IT department; it also changes the way government delivers its public services.”
Anyone who still regards AI as a matter for IT specialists alone is taking too narrow a view of what is happening, says Van Leeuwen. AI requires not only technical expertise and robust digital infrastructure but, above all, strategic leadership and governance.
This realisation prompted the Ministry of Justice and Security several years ago to move beyond treating AI solely as a CIO responsibility. Instead, it established a dedicated policy directorate responsible for AI, data protection and privacy.
AI is changing the landscape
“AI isn’t just changing the way government organisations work; it’s also changing the society they serve,” he explains. “That’s why AI is about much more than IT. We see this every day within the Ministry of Justice and Security. Think about cybercrime, deepfakes, copyright and privacy. These are not purely technical issues; they have direct implications for public policy, legislation and their implementation.”
According to Van Leeuwen, AI is reshaping government in 4 key ways. First, citizens, entrepreneurs and criminals alike are increasingly using AI. Cybercriminals, for example, use AI to make phishing campaigns more convincing or to identify vulnerabilities more quickly. Citizens, meanwhile, use AI to gather information or draft formal objections to government decisions.“That changes the environment in which government operates.”
Secondly, laws and regulations that have long been taken for granted need reconsideration. “Take copyright law. There are still court cases examining whether AI companies were entitled to use online content without permission to train their models. The resulting case law is highly relevant to our policy work.”
A broader public challenge
A third development is unfolding within government. Organisations are increasingly using AI to improve processes, support employees, and remain attractive employers. “That creates opportunities, but it also brings new responsibilities and requires new skills. We want and need to use AI responsibly. That’s why we support organisations across the Ministry in complying with existing legislation, including the European AI Act.”
Finally, citizens’ expectations are also changing. “People will soon be able to generate dozens of well-argued objections in minutes using AI. Governments cannot continue responding to all of these manually. People are becoming accustomed to using AI in their daily lives, and they also expect government to use AI to improve its own operations and services. That requires a different way of working.”
All of this demonstrates that AI affects not only internal operations but also the government’s wider public mission. “That’s why senior leaders cannot simply delegate this issue to the IT department.”
Start by understanding what you have
That does not mean leaders need to become AI experts. They do, however, need to understand which AI applications their organisation uses and the risks they may pose.
According to Van Leeuwen, responsible AI starts with a simple question: What AI systems and algorithms are we actually using? “Ask your CIO or Chief Data Officer to carry out an inventory. It sounds straightforward, but in practice it’s often a substantial exercise. Many organisations simply don’t know exactly which algorithms or AI applications they already have in place.”
Only then can organisations determine which systems fall within the scope of the AI Act and the risks they pose. This applies not only to new AI systems but also to existing algorithms. That is why, at the end of 2025, the Ministry of Justice and Security became the first Dutch government department to complete its algorithm register.
The AI Act is not a box-ticking exercise
For many organisations, the AI Act dominates discussions about AI. Van Leeuwen understands why but warns against treating it as merely a legal compliance exercise. “The AI Act encourages organisations to think carefully about how they use AI. Of course, compliance matters. But ultimately, it’s about being able to explain why you’ve made particular choices and ensuring that citizens can trust those decisions.”
He believes this is where senior leadership has a crucial role to play. “Don’t wait until every piece of guidance has been finalised. Invest now in knowledge, governance and awareness. Because many aspects of the AI Act still require further interpretation, it’s vital that organisations share their experiences and learn from one another. We support organisations with AI specialists, training and practical tools so that everyone doesn’t have to reinvent the wheel.”
“Can we?” vs “Should we?”
Van Leeuwen believes leaders should not approach AI solely through the lens of legislation and regulation. Organisations must, of course, comply with the AI Act and existing data protection laws. But an equally important question is what role AI should play in public service delivery. “The first question we often ask is: Can we do this? However, an equally important question is: Should we?”
This is where leadership responsibility lies. “Not everything that is technically possible or legally permissible is automatically appropriate for the public sector. Organisations will need to continually assess where AI genuinely adds value and where human judgement remains indispensable.”
Learning by doing
Anyone who believes government must have all the answers before using AI responsibly is mistaken, says Van Leeuwen. The technology is evolving far too quickly for that. “Experimentation isn’t a luxury; it’s essential if organisations are to build experience. You don’t learn about AI from a manual. Nobody learned how to use the internet by reading an instruction book first. You learn by using it.”
That does not mean every experiment has to succeed. “I only consider an experiment unsuccessful if you simply copy what your neighbour has done. Even if an application never reaches production, you’ve still learned how AI works, where its limitations lie, and where it does, or doesn’t, add value. The biggest challenge isn’t the technology itself; it’s embedding successful pilots into everyday practice. That means adapting your organisation, changing processes and bringing people with you.”
AI is more than chatbots
Within government, AI is also being used for pattern recognition, search functionality, software development and analysis of large volumes of data. “Many of these applications are less visible than chatbots, but they can deliver just as much public value.”
That also means organisations need to look beyond the latest trend. “Not every application requires generative AI, and not every challenge is solved by a chatbot. Ultimately, the question is always the same: where does AI genuinely add value within this particular process?”
Start today
What is the one message Van Leeuwen most wants to leave with leaders across government? His answer comes without hesitation. “I would encourage every leader to carry out their own AI experiment. Not because every organisation immediately needs a major AI programme, but because experience is the best teacher. Start small and explore where AI can genuinely make a difference.”
That brings him back to the central message of the interview: AI is neither a technological novelty nor an end in itself. “It’s a development that challenges organisations to rethink how they work, what their responsibilities are and how they create public value. We often overestimate what can be achieved in two years, but underestimate what is possible over ten.” Perhaps that is the most important lesson for public sector leaders, Van Leeuwen concludes. “Don’t wait until AI has fully matured. Start learning today.”
