In this investigation, I could show that the Swiss police uses a dubious software to assess the risk of individuals of being a “potential danger for the public”. While the use of the software had been publicly known until that point, I exclusively dug out some studies that show that the software actually performs not very well. In fact, out of three people it deems “potentially dangerous”, two actually weren’t.
The Swiss police and other state actors use more and more automated & algorithmic systems that take the burden of hard decisions off of them. I think that now is the right time to lay a special focus on such systems and investigate their (hidden) biases.
This story came together with an interactive simulation that explained the trade-off between false positives and false negatives.
This investigation won the 2019 Surveillance Studies price for journalists.