#CIJSummer

Across Europe, secretive algorithms are making life-changing decisions about millions of welfare recipients. An international team of journalists led by Lighthouse Reports gained unprecedented access to an AI system designed to predict which welfare recipients were committing fraud. This access enabled them to analyse the algorithm from the inside out and prove ethnic and gender discrimination. The result was a four-part series co-produced with WIRED magazine.
Lighthouse Reports’ lead reporter Gabriel Geiger takes us through the methodology of the two-year investigation, how they obtained the algorithm using public records laws, and the steps taken to run an academic-reviewed experiment to demonstrate bias.