Faster, better, safer – this is what the German government hopes to achieve with its proposed law on the use of artificial intelligence (AI) in migration administration. The draft, agreed upon by the government at the end of July, would affect not only asylum seekers but also individuals applying for visas in Germany or those granted temporary or permanent residence permits.

The law introduces three main changes: First, authorities dealing with asylum and residence laws would be allowed to use personal data to train or test AI systems. Second, they could establish “automated procedural monitoring” to identify patterns, such as whether applicants with similar characteristics are disproportionately rejected. Third, the Federal Office for Migration and Refugees (BAMF) and immigration authorities would be permitted to automatically cross-check data with online sources in cases of doubt about the information provided.

Compared to the initial draft that was leaked, the finalized government version includes minor improvements, such as explicitly prohibiting the use of biometric data for automated checks by BAMF and immigration authorities. However, these adjustments fall short of addressing the criticisms raised by various associations and federal commissioners.

Discrimination cannot be prevented by regulation alone

According to information from netzpolitik.org, the Federal Anti-Discrimination Commissioner, Ferda Ataman, expressed significant concerns during government consultations. Among other issues, she criticized the lack of sufficient discussion on banning discriminatory algorithms. The draft law merely states in several places that the relevant authorities must ensure that “discriminatory algorithms are neither developed nor used.” To achieve this, they are required to take the necessary “technical and organizational measures.” According to netzpolitik.org, Ataman called for annual transparency reports on the systems developed and deployed.

Several organizations have also criticized the insufficient safeguards against discrimination. Nora Oppermann from the NGO AlgorithmWatch warns against exposing people to “immature AI systems and using their fates for evidence-free research.”

Pro Asyl, in its statement, argues that the requirements for “system transparency, protection of highly sensitive personal data, and prevention of discriminatory decisions” are inadequate. Wiebke Judith, legal policy spokesperson for the human rights organization, warns in a press release: “A blank check for the application of AI across numerous authorities, as the draft law currently envisions, is extremely dangerous.” According to Judith, “This is about decisions that affect human lives.”

The organization calls on the German parliament to either abandon the draft law or fundamentally revise it. Parliamentary discussions on the draft are set to begin after the summer recess.

Efficiency must not come at the expense of constitutional principles

The Federal Commissioner for Data Protection and Freedom of Information (BfDI), Louisa Specht-Riemenschneider, also calls for revisions. A spokesperson for the agency told netzpolitik.org that while the agency recognizes that AI “can make administrative processes more efficient” and welcomes this goal, “efficiency must not come at the expense of constitutional principles.” From the BfDI’s perspective, improvements are needed during the parliamentary process.

The spokesperson added, “The draft law should clearly define which data may be used for the development and deployment of AI systems.” Where anonymized data suffices, personal data should not be used. If anonymized data is insufficient, pseudonymized information should be the next step. Only if these are inadequate should personal information be used for AI development and deployment.

Additionally, the automated matching of publicly available data constitutes “a significant intrusion into fundamental rights.” The spokesperson stated, “It is not enough to have the results of such matching reviewed by a human. It must first be verified whether there are concrete doubts justifying such a match. A purely algorithmic trigger must be excluded.”

Integration Commissioner welcomes AI plans

The Federal Integration Commissioner, Natalie Pawlik, views the initiative positively. “The development and use of AI in migration administration have the potential to reduce bottlenecks in application processing and shorten waiting times for decisions,” a spokesperson for the commissioner said. She welcomed, among other things, the “strict prohibition of discrimination,” which ensures that “the balance between technological innovation and data protection is maintained.”

In other countries that already use automated systems on a larger scale in migration administration, problems and potential pitfalls have become apparent: In 2023, journalistic investigations revealed issues and discrimination in the automated classification of visa applicants in the Netherlands. The system used categorized unmarried Nepalese men of certain ages as high-risk.

In a recent case from the United Kingdom, the Home Office relied on a document to reject an asylum application, the existence of which could not be definitively proven. A judge accused the ministry of basing its rejection on “AI-hallucinated” information. This underscores the importance of individual review for all applications.

The UK is also set to introduce new age recognition technology. Critics fear this could lead to even more unaccompanied minor refugees being classified as adults than is currently the case.

📝 The report was fully and exclusively transferred and translated by the Studies Department at AboMatrix

🔗 Source Document / المصدر الأصلي