Lighthouse Reports and Le Monde have investigated an algorithm used by France’s Caisse Nationale des Allocations Familiales (CNAF) to predict benefit fraud. The algorithm assigns a risk score to each of the 13 million households in France receiving benefits, with nearly seven in ten investigations being initiated based on a high risk score. The investigation revealed that the model disproportionately targets vulnerable individuals, such as single parents and those receiving disability benefits, potentially infringing on French anti-discrimination laws. The CNAF defended the model, claiming that it aims to classify files by complexity rather than establish a risk score. The agency has pledged to evaluate the model for bias and discrimination.
https://www.lighthousereports.com/methodology/how-we-investigated-frances-mass-profiling-machine/