Country
Full name of the document
Body / organism enacting the document
Nature of the body / organism
Type of document
Date
Project area
Law area
Full text of the document
Summary of the document
The document aims at mapping the human rights-related risks in using algorithms and to take measures to address this. Started out as a discussion document meant to answer questions and establish good decision-making practices for government organisations that consider developing, delegating the development of, buying, adjusting and/or using an algorithm. It identifies three stages through which a decision-making process concerning the development and implementation of an algorithm goes: 1) preparation, 2) input and throughput, 3) output, implementation and supervision. It provides a checklist for different questions to be answered in each phase with a focus on the expertise/role required for answering these questions (for example, commissioning client, project leader, domain expert, possibly citizen panel, possibly interest group representative). The authors hope that discussing the issues addressed in this document may contribute to algorithms being used in a careful, thoughtful, and well-embedded manner. For human rights it contains a roadmap of questions that the governmental organisation should address:
1. Fundamental right: does the algorithm affect (or threaten to affect) a fundamental right? 2. Specific legislation: does specific legislation apply with respect to the fundamental right that needs to be considered? 3. Defining seriousness: how seriously is this fundamental right infringed? 4. Objectives: what social, political, or administrative objectives are aimed at by using the algorithm? 5. Suitability: is using this specific algorithm a suitable tool to achieve these objectives? 6. Necessity and subsidiarity: is using this specific algorithm necessary to achieve this objective, and are there no other or mitigating measures available to do so? 7. Balancing and proportionality: at the end of the day, are the objectives sufficiently weighty to justify affecting fundamental rights?
Type of addressees
Territorial scope
Situations involved
Wherever government organizations are considering using or are using algorithms
AI system(s) involved
- All types of AI systems
Fundamental rights involved
- Freedom of expression
- Freedom of information
- Right to access to justice, to a fair trial and to jury trial
- Right to an effective remedy
- Right to data protection
- Right to good administration
- Right to privacy
- Right to non-discrimination
Principles expressly addressed
- Accountability
- Due process
- Effective (judicial) protection
- Equality
- Explainability
- Non-discrimination
- Proportionality
- Reasonableness
- Rule of law
- Transparency
Connection with hard law
Connected mainly to soft law and other policies:
● Integrated impact assessment framework for policy and legislation
● Algorithm assessment framework of the Netherlands Court of Audit (2021)
● Ethics Guidelines for Trustworthy Artificial Intelligence (2019)
● Guidelines for Algorithm Application by Governments and public education on data analyses
● An audit framework for algorithms (2020)
● Algorithm assessment framework Netherlands Court of Audit (2021)
● Good Digital Governance Code
● Guideline for public education on data analyses (2021)
● Data Protection Impact Assessment (DPIA)
● Non-discrimination by design guideline (2021)
● Government information security baseline (BIO)
● The FAIR and FACT-principles
Possible legal force and impact on national/supranational legal system
FRAIA found its way into the AI act of the EU and is likely to become hard law, possibly with some modifications, as soon as the AI Act enters into force (paragraph 5 of Article 29a of the joint text AI Act).