AI News
All stories
Reported

37 groups, including AlgorithmWatch, urge safeguards in Germany's AI migration bill

Ahead of a 23 November Bundestag hearing, 37 organisations call for binding evaluation, data protection and transparency rules for AI in migration cases.

Policy & Regulation · · 2 min read

Thirty-seven civil society organisations, with AlgorithmWatch in support, have called for changes to the KI-Migrationsverwaltungsgesetz (KIMVG) before its hearing in the interior committee on 23 November 2026. The bill would let migration authorities develop, train and use AI systems, and allow those systems to be trained on sensitive personal data. The appeal, dated 16 November 2026, says the bill grants wide powers without setting concrete rules for use, adequate protection of fundamental rights, clear responsibilities or sufficient transparency measures.

The appeal focuses on what the systems would do with data. According to the appeal, application data would be matched automatically against publicly accessible internet data. Patterns would be detected, and the results would support decisions on asylum, visas and residence. The organisations want personal data of affected people anonymised as a rule, written into law as binding. They also want whole asylum files and data stocks from the Ausländerzentralregister excluded from blanket use for AI training and automated analysis. Automated internet matching, they say, should face stricter conditions than the bill currently provides. The selection and oversight of external AI providers should follow clear statutory requirements.

On evaluation, the appeal asks for a fundamental rights impact assessment before the systems go into operation. Civil society should be involved, and the competent independent data protection authority should be included. The assessment should then continue while the systems run. The organisations also want use of AI in migration procedures documented in a traceable way, affected people informed, and effective ways to challenge decisions. They ask that the migration authorities' AI systems be entered in the government's "Marktplatz der KI-Möglichkeiten" and that evaluation results be submitted to the Bundestag.

AlgorithmWatch's own policy paper, written by Junior Policy Managerin Nora Oppermann and published on 15 October 2026, makes related arguments. It notes that the BAMF has used AI technologies for ten years. It argues that the new bill would extend automation in migration broadly, without the conditions needed for fair use. Its four recommendations: strike the planned changes to § 7 AsylG and § 86 AufenthG that permit AI development using personal data of people in the migration field; exclude passing personal data on to third parties; define effective measures against discriminatory algorithms; and create transparency and traceability for those affected and the public.

In a statement accompanying the appeal, Oppermann said the government was exposing refugees to high AI risks in asylum and visa procedures, and was apparently accepting all consequences. She wrote that processing highly sensitive data with AI always carries a risk of discrimination and wrong decisions. For people fleeing, she said, this could mean a concrete threat to life. She called the approach of trying things out and seeing what happens absolutely unacceptable in migration procedures. She also warned that the law could become a blueprint for even more surveillance of everyone.

What changed

Type: Rules and regulation · Applies to: KI-Migrationsverwaltungsgesetz (KIMVG)

Why it matters

If passed in its current form, the bill would give German migration authorities broad powers to use AI on sensitive case data, and the rules for choosing and supervising external AI vendors would be set by law.

Original source

Where we saw it

1 source

AlgorithmWatch

AI wrote this summary from the sources listed, and nobody reviewed it individually before publication. Check the original source before you rely on it. How it works