Research

Bureaucratic Silences: What the Canadian AI Register Reveals, Omits, and Obscures (opens in a new tab)

Das, D., Tessono, C., Ahmed, S. I., & Guha, S. · arXiv preprint 2604.15514 (submitted 16 April 2026) · 2026

Across all 409 systems in Canada's federal register, 86% are internal efficiency tools, and the register consistently leaves out human discretion, training and how uncertainty is handled.

A Global Comparison of Schemas, Transparency, and Interoperability in Public-Sector AI Registers and Inventories (opens in a new tab)

Das, D. & Guha, S. · arXiv preprint 2609.24883 (submitted 21 September 2026); ACM DOI listed on the arXiv page · 2026

Comparing 8,368 records from 17 instruments covering 72 countries, registers share a descriptive core but rarely ask about appeals, risks, legal bases or external evaluation.

Are Algorithm Registers Transparent? Perspectives from Germany (opens in a new tab)

Peljto, I., Heilmann, X., & Cerrato, M. · arXiv preprint 2606.02347 (v2, 2 September 2026) · 2026

Turns a proposal for a national German register into an audit checklist, and finds that both main German platforms need substantial changes before they work as transparency tools.

Co-constructing sociotechnical AI governance: participatory system mapping using algorithm registers (opens in a new tab)

de Troya, I., Enbergs, M., Doorn, N., & Dobbe, R. · arXiv preprint 2608.12166 · 2026

Combining a Dutch municipal register entry for a rule-based welfare-eligibility tool with workshops (N=8) surfaced hazards the register alone did not show, such as wrongful denial of benefits and no way to contest decisions.

Making Algorithm Registers Work for Meaningful Transparency (opens in a new tab)

Pénicaud, S. · IA Ciudadana report, March 2025 (coordinated and edited by IA Ciudadana) · 2025

Most registers are hard to find and mostly list systems that are not high-risk under the AI Act; the report gives recommendations on scope, discoverability and lifecycle coverage.

Algorithmic Transparency in the Public Sector: A state-of-the-art report of algorithmic transparency instruments (opens in a new tab)

Gutiérrez, J. D. & Muñoz-Cadena, S. · Global Partnership on AI (GPAI), May 2024; hosted on OECD.AI · 2024

Inventories registers and other transparency instruments across countries, and notes that items such as system metrics and justification were often not found.

Dutch Comfort: The Limits of AI Governance through Municipal Registers (opens in a new tab)

Cath, C. & Jansen, F. · Techné: Research in Philosophy and Technology 26(3):395-412 · 2022

The Amsterdam and Helsinki registers mostly list uncontentious uses and leave out sensitive systems such as welfare fraud detection, so treating them as accountability risks ethics theatre.

Accountable Artificial Intelligence: Holding Algorithms to Account (opens in a new tab)

Busuioc, M. · Public Administration Review 81(5):825-836 · 2021

A public administration analysis of how opacity, proprietary secrecy and expertise gaps weaken accountability for AI in government.

Algorithmic Accountability for the Public Sector: Learning from the First Wave of Policy Implementation (opens in a new tab)

Ada Lovelace Institute, AI Now Institute, & Open Government Partnership · Joint report, published 24 August 2021 · 2021

A comparative review of registers, impact assessments, audits and bans, drawing on the experience of the people who implemented them; context and public participation decide whether these tools work.

What to account for when accounting for algorithms: a systematic literature review on algorithmic accountability (opens in a new tab)

Wieringa, M. · Proceedings of FAT* '20 (ACM), pp. 1-18 · 2020

A systematic review that applies Bovens' accountability framework (actor, forum, account, consequences) to algorithmic systems.

Seeing without knowing: Limitations of the transparency ideal and its application to algorithmic accountability (opens in a new tab)

Ananny, M. & Crawford, K. · New Media & Society 20(3):973-989 · 2018

Being able to see a system is not the same as understanding or governing it; the authors set out limits of the transparency ideal and argue for accountability across whole sociotechnical systems.

DUO student-grant checks (Netherlands)

Auditing a Dutch Public Sector Risk Profiling Algorithm Using an Unsupervised Bias Detection Tool (opens in a new tab)

Holstege, F., Jorgensen, M., Padh, K., Parie, J., Proroković, K., Persson, J., & Snoek, L. · Proceedings of the IASEAI Conference 2(1):240-253 (IASEAI'26), published 15 July 2026; arXiv:2502.01713 · 2026

Re-audits DUO's student-grant risk profile without individual demographic data. Unsupervised clustering flags vocational students and students living near their parents, which Statistics Netherlands data show to be proxies for non-European migration background. An open-source tool is released.

Barcelona Activa public employment agency (Spain)

Applied and Filtered: An End-to-End Algorithmic Fairness Audit of a Public Employment Agency (opens in a new tab)

Galdón-Clavell, G. · arXiv preprint 2608.13022 (FINDHR project, Horizon Europe) · 2026

Audits about 497,000 candidate-vacancy records (2017-2022). Overall gender parity hides disparities by salary, age and gender identity, and fairness conclusions change over time; a model-level audit alone would have missed these.

Rotterdam welfare risk model (Netherlands)

Algoprudence: Risk Profiling for Social Welfare Re-examination (ALGO:AA:2023:02) (opens in a new tab)

Algorithm Audit · Algorithm Audit (NGO), report with an advisory commission · 2023

A deliberative review by an expert commission of the Rotterdam model's variables, obtained through public-records requests, flagging proxy-discrimination risks such as 'illiteracy' and subjective traits.