Commission cites discrimination risks in data-driven fraud profiling, urges a halt
A Dutch state advisory body is urging the government to stop using large-scale data analysis to profile citizens in efforts to detect fraud and crime, warning that the practice risks structural discrimination and undermines the rule of law, Trouw reports.
The call follows years of controversy, including the childcare benefits scandal and a 2023 court ruling in The Hague on ethnic profiling by border police.
The State Commission against Discrimination and Racism said Wednesday it recommends a temporary halt to so-called “data-driven profiling” in fraud and crime enforcement until safeguards are in place. The advisory body was created in part in response to the childcare benefits affair.
The commission cited what it called an “explosive growth” in government datasets that are increasingly linked together to assess citizens’ risk profiles. These systems are used in address investigations, neighborhood-based enforcement, and fraud detection related to benefits and allowances.
As a result, groups of citizens are selected on the basis of statistical risk indicators. The commission said this can lead to discrimination in practice, even when the stated goal is fraud prevention.
Past problems include recipients of childcare benefits with dual nationality and students with a migration background who were wrongly flagged in connection with housing-related student grants. “Citizens often do not know that they are being profiled, what criteria are used, and how decisions are made,” the commission said.
Joyce Sylvester, chair of the State Commission against Discrimination and Racism, stressed that the government must be transparent about its use of profiling tools.
“Because profiling touches on core principles of the rule of law, democracy, and good governance, the government must be transparent about its choices and demonstrate that the use of profiling is effective and does not cause harm,” Sylvester told Trouw.
The commission said oversight bodies, including the Dutch Data Protection Authority, have repeatedly warned in recent years about unlawful or discriminatory processing of personal data. It also noted that monitoring algorithmic systems and artificial intelligence has become increasingly difficult as their use expands across government agencies.
Human rights organizations have for years described the Netherlands as developing into a “prevention state,” in which citizens are assessed not only on their actions but also on what they might do based on statistical indicators.
The Scientific Advisory Council for Police reported last year that Dutch police work is increasingly guided by large datasets, which are used to improve what officials describe as an “optimal information position.”
The commission also criticized what it sees as a lack of sustained political debate once discriminatory effects are identified. It said discussions tend to focus on isolated incidents rather than on whether profiling itself is an appropriate general method.
According to the commission, the government has not taken earlier criticism seriously enough. It pointed to a 2023 ruling by the Court of Appeal in The Hague, which found that the Royal Netherlands Marechaussee border police may not use skin color as a basis for identity checks. The government later concluded that the ruling applied only to that specific case and did not affect broader data-driven profiling practices.
The commission acknowledged that stopping data-driven profiling would significantly affect government agencies and implementation bodies. However, it warned that without strong safeguards, continued use of such systems carries “a considerable risk of lasting social damage and high recovery costs.”
