Gergely Csurgai-Horváth: Regulating algorithmic bias as a key element of digital market regulation. Nr. 2024/1.

This paper addresses the rules applicable to algorithmic bias taking the form of self-favouring by hybrid digital platforms in the EU. In this paper, it is argued that the recently introduced prohibition of self-favouring by digital platforms should not apply across the board in the same manner. It may be necessary to consider the nature of the underlying products or services, the business models, and the monetisation strategies of digital platforms. Differences in these aspects may alter their ability and incentives to engage in self-favouring potentially leading to foreclosing rivals and harming consumers. This suggests that the approach put forward by Section 19a of the German Competition Act may be better from an error-cost perspective than that of the Digital Markets Act. Section 19a of the GWB grants more discretion to enforcers and allows for a broader justification of the impugned conduct. In the context of the DMA, some sort of balancing exercise seems to be possible only if the European Commission makes extensive use of the possibility to further specify the prohibition of self-favouring contained in Article 6(5) of the DMA in light of the principles of effectiveness and proportionality. Finally, the paper touches upon the potential disproportionate burden, legal fragmentation, and legal uncertainty across the EU resulting from the interplay between EU competition law, the DMA, and national laws tackling similar self-favouring practices.

Keywords: Algorithmic bias, Algorithmic foreclosure, Self-favouring, Digital platforms, Digital Markets Act, Competition law, Economic market regulation, Platform to Business Regulation, Section 19a of the German Competition Act

The working paper is available for download here.