Use of artificial intelligence and social scoring by Danish social security authority

Priority question for written answer P-002847/2024
to the Commission
Rule 144
Maria Ohisalo (Verts/ALE), Villy Søvndal (Verts/ALE)

The Danish social security authority (UDK) is using artificial intelligence (AI) to detect social benefits abuses. Reportedly, the system uses social scoring that risks discriminating against certain groups of people and has enabled digital mass surveillance[1].

The purpose of the EU AI Act is to enable human-centric and trustworthy AI, while ensuring the protection of fundamental rights. While the AI Act prohibits 'social scoring', the definition remains unclear.

• 1.In the Commission's view with regard to the AI Act, is the system used by the UDK in line with the obligations of the AI Act, and on what conditions would the system used by UDK be prohibited under the AI Act[2]?
• 2.In the Commission's view, has the UDK undertaken an adequate impact assessment of the system in relation to fundamental rights, as required by the AI Act[3], and has the UDK ensured robust transparency and adequate oversight of the use of its algorithm?
• 3.How will the Commission clarify which practices count as social scoring and provide guidance on the practical implementation of the prohibited practices referred to in the AI Act, particularly in the social services sector?

Submitted: 10.12.2024

• [1] https://www.amnesty.org/en/documents/eur18/8709/2024/en/.
• [2] Article 5.1.
• [3] Article 27.