Research Summary: Are Algorithm Registers Transparent? Perspectives from Germany

Research from arXiv highlights that algorithm registers, designed to provide transparency into public sector algorithms, face challenges in achieving meaningful transparency. Specifically in Germany, the landscape is fragmented with no unified federal register, leading to varying scopes and objectives across existing state and federal initiatives. A conceptual proposal from 2025 outlines requirements for a national AI transparency register, which this research uses as an audit instrument.

Why it matters

The transparency of algorithms used in public administration is crucial for public trust, accountability, and democratic oversight. Fragmentation in regulatory approaches and disclosure standards can lead to inconsistencies, hinder public scrutiny, and potentially undermine the legitimacy of algorithmic decision-making.

Key insights

• Algorithm registers are public databases intended to disclose information about algorithms used in public administration.
• The effectiveness of these registers in delivering meaningful transparency is a subject of ongoing debate.
• Germany lacks a consolidated federal algorithm register, resulting in a fragmented landscape of at least five distinct state and federal initiatives.
• These initiatives operate with diverse scopes and objectives regarding AI system information disclosure.
• A conceptual proposal from 2025 by Alina Lorenz outlines technical and governance requirements for a national AI transparency register in Germany.
• This research repurposes Lorenz's proposal as an audit instrument to evaluate existing transparency efforts.

Source

arXiv — Computers and Society — https://arxiv.org/abs/2606.02347