The Immigration and Naturalization Service has been using two artificial intelligence tools to assist staff with asylum applications since this summer.
Internal documents show the systems, named Robin and ChatIND, were never entered into the Dutch national algorithm registry or the European register.
Preparatory tasks for staff
Staff use Robin as an assistant to ask questions and assign tasks like a colleague. ChatIND translates texts, summarizes long files, and analyzes interviews and research documents.
An agency spokesperson stated that the tools only handle preparatory work and staff are barred from letting algorithms make final asylum decisions.
The agency claimed registration was unnecessary because an internal assessment found the software posed no high risk.
That decision directly contradicts the agency’s published AI policy, which promises to register all algorithmic systems regardless of risk level.
Lawyers raise fairness concerns
Researchers and legal professionals have raised objections over how the software might sway asylum outcomes.
Ola al Khatib, an algorithm researcher at Utrecht University, noted that staff often place too much trust in AI answers, meaning errors can seep into final rulings.
Wil Eikelboom of the Association of Asylum Lawyers said summarizing and analyzing files forms a central part of decision making that courts must be able to review.
The legal association expressed deep concern that the systems were introduced without notifying lawyers representing asylum seekers.
