evidoria

← Back to browse

Good practice Imported

Rekenkamer Metropool Amsterdam's Algorithm Audit — Amsterdam's Framework Covered 49 of 61 Risks, Yet Only One Algorithm Was Tested for Fairness

Netherlands · Amsterdam · See the Netherlands profile · See the Amsterdam profile

Top 47% 60/100 · Ask Evidence Copilot about this practice

In October 2023 the Amsterdam Court of Audit tested the city's algorithm governance: its framework addressed 49 of 61 risks, but fairness was examined for only one algorithm and the register held 29 entries.

Rekenkamer Metropool Amsterdam's Algorithm Audit — Amsterdam's Framework Covered 49 of 61 Risks, Yet Only One Algorithm Was Tested for Fairness

Details

Promoter
Rekenkamer Metropool Amsterdam (Amsterdam-Zaanstad Court of Audit)
Period
2023
Keywords
algorithm audit, algorithm register, municipal governance

Description

In October 2023 the Rekenkamer Metropool Amsterdam published 'Algoritmen', an independent audit of the guidelines Amsterdam set to make its algorithms transparent and responsible. It examined three algorithms: a decision-support system for self-employed benefit applications (Bbz) and two tools that flag potential water-related and social-housing violations.
The audit found the municipal framework addressed 49 of 61 identified risks across governance, data and modelling, privacy, IT management and ethics. It concluded that the city had insufficient grip on its algorithms: fairness had been thoroughly investigated for only one algorithm, legal privacy checks were still insufficient, citizens were not clearly told how algorithms were used in decisions about them, and the organisation had less visibility than it had estimated — the municipality believed it used hundreds of algorithms but only 29 appeared in the public register in October 2023.
The council accepted all five recommendations on 20 December 2023 (stronger fairness and privacy safeguards, a more consistent framework, a fuller register, clearer communication to citizens) and requested an implementation plan. The case is a cautionary example: a pioneer of algorithm registers shows that publishing a register does not by itself guarantee control, and that independent audit of the practice is needed. No outcome data on the follow-up are cited here.

Read the full analysis: https://www.rekenkamer.amsterdam.nl/onderzoek/algoritmen/

Implementation

Implementation detail (cost, timeline, staffing, conditions for success) is not yet available for this practice.

Do you run this practice? Claim it — verified implementers get a public contact pathway and can propose corrections.

Data sources

Where this practice's information was retrieved from, and when.

Attachments

Similar practices you may find useful