Denmark’s AI-Powered Welfare System: A Double-Edged Sword
The Danish welfare authority, Udbetaling Danmark (UDK), is facing scrutiny for its use of artificial intelligence (AI) tools that may inadvertently discriminate against vulnerable populations. A recent report by Amnesty International has raised concerns about the implications of these technologies, particularly how they flag individuals for social benefits fraud investigations.
Mass Surveillance and Privacy Erosion
Amnesty’s report, titled Coded Injustice: Surveillance and Discrimination in Denmark’s Automated Welfare State, outlines a troubling reality where the extensive use of fraud detection algorithms, combined with mass surveillance practices, compromises individual privacy. The report highlights how individuals are often forced to relinquish their rights without their knowledge, creating an atmosphere of fear among those who depend on social benefits.
According to Hellen Mukiri-Smith, Amnesty International’s Researcher on Artificial Intelligence and Human Rights, “This mass surveillance has created a social benefits system that risks targeting, rather than supporting the very people it was meant to protect.”
Algorithmic Models and Data Collection
UDK has employed a range of up to 60 algorithmic models to detect social benefits fraud. These models are designed to flag individuals for further investigations, leading to the potential for discriminatory practices. The algorithms draw on a vast pool of personal data collected from public databases, which includes sensitive information such as residency status, citizenship, and family relationships. This data can serve as proxies for a person’s race, ethnicity, or sexual orientation.
UDK maintains that this extensive data collection is “legally grounded,” but Amnesty International argues that the scale of data processed is neither necessary nor proportionate to the goal of fraud detection.
The Psychological Toll on Individuals
Individuals subjected to surveillance have reported severe psychological impacts, including anxiety and depression. Gitte Nielsen, Chairperson of the Social and Labor Market Policy Committee at Dansk Handicap Foundation, noted that people with disabilities often feel “interrogated” by case workers, leading to feelings of despair.
One interviewee described the experience as “sitting at the end of the gun,” indicating a continuous state of fear and anxiety regarding potential scrutiny by authorities.
Discrimination Through Algorithms
The algorithms implemented by UDK and its partner, Arbejdsmarkedets Tillægspension (ATP), perpetuate existing inequalities in Danish society. For instance, the Really Single algorithm aims to predict a person’s family or relationship status but lacks clarity on what constitutes “unusual” living arrangements. This ambiguity may lead to arbitrary decision-making that disproportionately affects marginalized groups.
Moreover, the Model Abroad algorithm targets beneficiaries with “medium and high-strength ties” to non-EEA countries, raising concerns about discrimination based on national origin and migration status. Amnesty International asserts that the use of “citizenship” as a parameter in these algorithms can indirectly reveal sensitive information about an individual’s race and ethnicity.
Calls for Transparency and Regulation
Amnesty International has urged the European Commission to provide clearer guidelines in the EU Artificial Intelligence Act regarding what constitutes social scoring, particularly in the context of fraud detection. The organization stresses the need for robust transparency and oversight in the development and deployment of these algorithms.
“The Danish authorities must urgently implement a clear and legally binding ban on the use of data related to ‘foreign affiliation’ or proxy data in risk-scoring for fraud control purposes,” Mukiri-Smith stated, underscoring the importance of safeguarding human rights.
Conclusion
Denmark’s approach to AI in the welfare system highlights a critical intersection between technology and human rights. As UDK and ATP utilize sophisticated algorithms to manage social benefits, the potential for discrimination and privacy erosion raises pressing ethical questions. Ensuring that these systems support rather than undermine individual dignity and rights will be paramount as society navigates the complexities of automated decision-making.
