Amnesty International reports Denmark's welfare AI algorithms risk discriminating against vulnerable groups
Amnesty International reported that Denmark's welfare fraud detection algorithms, deployed by Udbetalning Danmark, risk discriminating against vulnerable populations by using citizenship and other sensitive personal data as proxies for protected characteristics. The organization warned that the system creates mass surveillance that may target rather than support the people it was designed to help.
What happened
Amnesty International examined four algorithms used by Udbetalning Danmark, Denmark's social benefits agency, among approximately 60 algorithms deployed over more than a decade to identify welfare fraud. The algorithms process personal data including place of residence, travel, citizenship, place of birth, family relationships and income from public databases of Danish residents.
One algorithm, 'Model Abroad,' is based on the beneficiary's nationality to determine whether people have moved abroad while still receiving social benefits. Amnesty argued that using citizenship as a primary criterion violates non-discrimination rights.
Reported impact
Amnesty International stated that mass surveillance through these algorithms has created a system that risks targeting rather than supporting vulnerable populations including people with disabilities, low-income families, and migrants. The organization warned the system is creating an atmosphere of fear among welfare recipients.
Index analysis
The algorithms use sensitive data points—such as citizenship, place of residence and family relationships—that Amnesty noted can serve as proxies for race, ethnicity, or sexual orientation. Amnesty argued that feeding personal data into algorithmic models to identify fraud risk requires greater care to avoid discrimination.
Classification
Sources and evidence
This record was researched and written by the Index, based on the following original reporting.