Fundamental rights impact assessment - a practical tool for protecting fundamental rights

• Author: European Union Agency for Fundamental Rights (EU body or agency)
• Pages: 87-100





FUNDAMENTAL RIGHTS IMPACT

ASSESSMENT APRACTICAL TOOL

FOR PROTECTING FUNDAMENTAL

RIGHTS

Chapter 4 illustrated the extent to which using AI affects different fundamental

rights. This chapter analyses how fundamental rights impact assessments

(FRIA) could reduce the negative impacts that using AI can have on

fundamental rights.

Section 5.1 provides a brief overview of the current discussion on the need

for fundamental rights impact assessments in this f‌ield. Section 5.2 analyses

current practices in addressing fundamental rights implications, based on the

interviews conducted for this report. Interviewees were asked about what

sort of testing was done before the system was used, and who controls the

tasks affected by the use of the technology.

The chapter ends with suggestions on how to assess the fundamental rights

impact when using AI and related technologies.

.. CALLING FOR AFUNDAMENTAL RIGHTS IMPACT

ASSESSMENT AVAILABLE GUIDANCE AND TOOLS

International organisations,

1

academics

2

and civil society

3

have called for

fundamental rights impact assessments to be conducted when using AI or

related technologies.

For example, the Committee of Ministers of the Council of Europe’s guidelines

on addressing the human rights impacts of algorithmic systems recommend

that states should conduct “impact assessments prior to public procurement,

during development, at regular milestones, and throughout their context-

specif‌ic deployment in order to identify the risks of rights-adverse outcomes”.

4

There is a need for f‌lexible impact assessments that can adapt to different

situations given that fundamental rights violations are always contextual.

Scholars exemplify this based on EU anti-discrimination law, where equality

is always contextual and depends on the case at hand.5

Fundamental rights compliance cannot be automated and hard-coded into

computer software. Rather, each use case needs separate examination

to determine whether any fundamental rights issue arises. Nevertheless,

assessments can follow a systematic approach and provide similar information.

Existing standards provide guidance on how to do a fundamental rights impact

assessment of AI and related technology. These include hard law, soft law