Obscure government algorithms are making life-changing decisions about millions of people around the world. Now, for the first time, we can reveal how one of these systems works. For Suspicion Machine, a four-part series from WIRED and Lighthouse Reports, we gained unprecedented access to one of the world's most sophisticated welfare fraud detection algorithms.
We obtained not only the algorithm itself, but also the data that powered it and the handbook used by the data scientists who ran it. This allowed us to see how such a sophisticated system evaluated people as potentially committing benefits fraud. What we found was deeply concerning.
The system, used by the city of Rotterdam in the Netherlands until 2021, discriminated against people based on their ethnicity and gender. We also found evidence of fundamental flaws that made the system inaccurate and unfair.
From the length of your last romantic relationship to how often you play sports, risk-scoring algorithms combine invasive and banal data points to decide whether you are likely to have committed welfare fraud—and whether your benefits should be taken away. Our reporting allows you to see how the algorithm works for yourself, and to experiment with its risk-ranking system to understand how it discriminates against certain people based on characteristics they have no control over.
In our first story, we show you how algorithms discriminate against people based on their age and ethnicity. In our second story, we reveal the human impact of biased risk-scoring algorithms. Our third story interrogates the politics that has led to the rise of these broken systems. Finally, in our fourth story, we show how a combination of secretive governments and yet more secretive private companies has created a system in which lives are ruined—with little hope of justice.
What we found doesn't just affect people living in Rotterdam. Risk-scoring algorithms are commonplace around the world, yet there has been little transparency around how they work—until now. Read all the stories in our four-part series and delve deep inside the Suspicion Machine.
