AI lessons from the boardroom
Inside the Police’s AI approach to environmental crime
Police innovator Amir Niknam on environmental crime, complexity and the limits of technology
By Rutger Huizenga
Executives are increasingly confronted with AI. Not as a gadget, but as a governance question: what should you use it for, who remains responsible, and what should you not use it for? In the new interview series AI Lessons from the Boardroom, we speak with people who have to make those decisions in practice. Today: Amir Niknam, senior innovation advisor at the Dutch National Police, about how AI helps him combat environmental crime.
Niknam is not a typical technology executive. He studied psychology, worked in IT for many years and started at the police as a detective. Since last year he has been part of the High Impact Environmental Crimes team, which deals with the most serious and complex environmental cases in the Netherlands and internationally. These range from illegal waste dumping on industrial sites to cases involving international supply chains, permits, chemical substances and major financial interests.
Environmental crimes are close to Niknam’s heart. He even supported the Extinction Rebellion demonstrations against fossil fuel subsidies that blocked the A12 motorway, of course without breaking any rules or hindering colleagues on duty. Environmental safety, he says, is not activism but an urgent security issue.
“The damage is less visible than street violence, but the impact on health and living conditions is enormous. Some Dutch cities demonstrably face years shorter life expectancy because of air pollution. And we need to stop that.”
Playing legal Champions League
Environmental cases are notoriously difficult. A single investigation often touches simultaneously on criminal law, administrative law, permits and European regulations. Money and goods flow through producers, transporters and traders, often across borders.
“Anyone who wants to intervene here has to play in the legal Champions League”, says Niknam. “Take the wrong approach and you’re stuck in endless procedures. And time tends to favor companies with deep pockets and strong legal teams. Complexity can become a shield.”
The bottleneck often appears at the start of a case. There may be a signal or report, but what is the most promising legal route? That first analysis currently takes detectives weeks or even months, while the damage continues.
"Al Capone" as a principle
This is exactly where Niknam and KickstartAI are focusing with an AI tool currently under development. The project, winner of the National AI Challenge 2025, has internally been nicknamed Al Capone.
The reference is telling: the gangster was not convicted of murder, but of tax evasion.
“You see that principle in environmental cases as well”, Niknam explains. “The most serious violation is not always the easiest to prove. Sometimes it’s more effective to revoke a permit or shut down a company than to wait for a drawn-out criminal trial.”
The AI tool only supports the initial analysis phase. An intelligence officer enters context, a report, an anomaly, a suspicion. The system searches accessible legislation, case law and similar cases, but it does not issue judgments.
“You want to see more quickly where the legal opening might be. After that, it remains human work.”
Niknam stresses that he is not responsible for the overall AI policy of the police. “I’m speaking here in a personal capacity.” Yet the questions he encounters, about responsibility, boundaries and legitimacy, are broader ones. Within the police, AI is not introduced lightly, he says.
“We operate in a constitutional state. That requires explainability, proportionality and human judgment.”
It also means that not everything that is technically possible is automatically desirable.
“The police are vigilant and serve the values of the democratic rule of law. That is our mission, and it must always come first.”
The broader lesson
This is also the lesson he wants to share with other organizations. “Never let AI become a strategy in itself. It should always be a means.”
In his own work, the starting point was very concrete: the slow start of complex environmental investigations. Not reforming the entire police organization, but addressing one clear bottleneck.
According to him, that’s where things often go wrong. Organizations start with the technology instead of the friction.
“The danger is that new technology ends up reinforcing the very problem you’re trying to solve”, he says.
He also sees how many organizations are still structured linearly, according to the model of the industrial revolution: step one, then step two, then step three. That worked in factories, but in knowledge work it leads to waiting times, handovers and layers of control.
“You end up with gatekeepers everywhere. Another check, another form. And the system gradually clogs up.”
The temptation is to replace those gatekeepers with AI. “But that’s like buying a better mop while the tap is still running.”
Building blocks instead of chains
According to Niknam, the real intervention lies in the design of the process itself. Why must work constantly be handed over? Why not organize it more asynchronously? He advocates working with modular building blocks, like a house where you can renovate one room without demolishing the entire building.
In such a system, AI is a tool, not a load-bearing wall. “AI must be replaceable. Switchable off when the context changes. If you can’t disconnect it without everything collapsing, you have a problem.
Because if the structure itself is flawed, AI won’t fix it. “It will only make things go wrong faster.”
“Don’t wait because the protocol says so”. Niknam concludes. “Act as soon as you can. And that’s exactly where AI can be a wonderful tool.”
