Skip to content
TechTimes·Indexed by this site · Originally published· Governance →· Its place in the scan record →

Germany's draft AI immigration management law relies on staff training; a research review says training cannot prevent automation bias

Germany AI Asylum Law Draft Leans on Staff Training That Review Says Cannot Stop Automation Bias

AI summary

Germany's draft AI immigration management law (KIMVG) relies on human review and staff training to prevent over-trust in machine outputs. However, a 2010 Human Factors review found that automation bias occurs among both novices and experts and cannot be prevented through training or instructions, and the draft's full text does not use the term automation bias. The draft would allow the Federal Office for Migration and Refugees (BAMF) and foreigners authorities to use case data to train, validate and test AI systems, to conduct procedural monitoring of case processing, and to compare applicants against publicly available information on the internet, with no general anonymization obligation. According to netzpolitik, on Monday 37 civil society organizations including AlgorithmWatch called on the Bundestag to amend the draft before the public hearing of the Interior Committee on November 23. The EU AI Act classifies AI that assists in asylum, visa and residence assessments as high-risk under Annex III, but Regulation (EU) 2026/1744 has postponed the relevant obligations to December 2, 2027.

Read originaltechtimes.com