Generative AI in Check: Gatekeeper Power and Policy Under the DMA

Abstract

The deployment of Artificial Intelligence systems has become mainstream among small and large economic operators. One of the most significant developments includes services that rely on natural language processing. Multimodal AI and generative AI systems have attracted regulatory interest due to their global adoption. Concentration levels at different levels of distribution of these systems have seen a particularly stark increase. According to regulators, such a trend prompts the need to intervene so that AI systems can remain transparent, explainable, non-discriminatory, and fair.

In the EU, enforcers of the Digital Markets Act are already considering how to square the circle of capturing AI through its provisions. This paper identifies a challenge to such an integration: the issue of scope. To design effective regulatory reforms aimed at addressing concentration, it is essential to clarify which aspects of AI systems should be targeted. In line with the DMA’s spirit, the paper argues that legislative developments should take place at the input level as well as regarding training data. 

Keywords: Artificial Intelligence, Generative AI, Digital Markets Act, Regulatory Capture