CAF's fraud detection algorithm accused of discrimination
A coalition of civil society organizations reportedly filed a legal challenge in France seeking to abolish the CAF's fraud detection algorithm, alleging it discriminates against economically disadvantaged beneficiaries and violates data protection rules. The algorithm has been used since 2010 to score allocatees for targeted audits based on personal data.
What happened
A coalition of approximately 15 civil society organizations, including La Quadrature du Net and Amnesty International, filed a legal challenge with the French administrative court (Conseil d'État) on October 16, 2024, seeking to abolish the CAF's fraud detection algorithm.
The algorithm has been in use by the Caisse nationale des allocations familiales (CAF) since 2010 to score and assign allocatees to control audits based on personal data. It assigns a score from 0 to 10 to households using factors such as income below a threshold, presence of a child over 12, and recent widowhood or divorce.
Reported impact
The organizations allege that the algorithm discriminates against the most economically precarious allocatees by targeting them disproportionately for controls and claims for overpayment.
The algorithm processes personal data of 32 million people, and the organizations question whether this volume of data processing is proportionate to the stated objective.
Index analysis
The organizations argue the algorithm lacks transparency, with the CAF rarely explaining the calculations or reasoning for control decisions.
The plaintiffs contend the algorithm's use may violate the General Data Protection Regulation (GDPR) due to the scope and disproportionality of data processing relative to its purpose.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.