Government Risk Profiling: Discrimination Risks Remain High

Margriet Vermeer · 2026-05-07

Government agencies use risk profiling to decide who gets flagged for extra scrutiny. It sounds clinical, even necessary. But without strict safeguards, these systems can cause real harm. A new report from the Universiteit van Amsterdam warns that risk profiling, when done without principled measures, keeps the door open to discrimination. The researchers argue that we need more than just good intentions. We need rules that actually protect people. ### What Is Risk Profiling, Exactly? Risk profiling is when the government uses data to predict behavior. Think of it like a credit score, but for security or law enforcement. They look at patterns and assign a risk level to individuals or groups. Sounds efficient, right? The problem is that these systems often rely on biased data. If past policing targeted certain neighborhoods, the algorithm learns to flag those same areas again. It creates a loop that keeps hurting the same communities.  ### The Core Problem: Bias Built Into the System The Universiteit van Amsterdam study highlights a few key issues: - **Data bias**: Historical data reflects past discrimination. Using it to train algorithms just repeats those mistakes. - **Lack of transparency**: Most profiling systems are black boxes. People don't know why they were flagged, and they have no way to challenge it. - **No accountability**: When a system makes a mistake, who is responsible? The developer? The agency? Usually, no one. The report makes it clear: without principled measures, the risk of discrimination and harm remains too great. ### What "Principled Measures" Look Like The researchers aren’t saying we should scrap all risk profiling. They’re saying we need to do it right. Here’s what that means: - **Independent oversight**: An outside body should review every profiling system before it goes live. - **Regular audits**: Systems need constant checking for bias, not just a one-time review. - **Right to explanation**: If you’re flagged, you deserve to know why. And you should be able to appeal. - **Data minimization**: Only collect what’s absolutely necessary. The less data, the less room for abuse. > "Without principled measures, the risk of discrimination and harm remains too great." — Universiteit van Amsterdam researchers ### Why This Matters Right Now Risk profiling isn’t some future tech. It’s happening today. From airport security to welfare fraud detection, algorithms are making decisions that affect real lives. In the United States, similar concerns have been raised about predictive policing and credit scoring. The same patterns emerge: biased data, lack of transparency, and no real accountability. ### A Path Forward The report doesn’t just point out problems. It offers solutions. The key is to build fairness into the system from the start. That means involving communities, testing for bias, and giving people real power to challenge decisions. We don’t have to choose between safety and fairness. With the right rules, we can have both. ### Final Thoughts Risk profiling is a tool. Like any tool, it depends on how you use it. Used carelessly, it amplifies existing inequalities. Used thoughtfully, with principled measures, it can be part of a fair system. The question is whether governments will take that path. The evidence says they need to start now.