Regulators Must Think Beyond Siloed Markets to Tame Agentic AI Concentration
Open Markets Institute Europe was back at the Computers, Privacy and Data Protection Conference (CPDP) in Brussels in May, hosting a panel on power concentration in agentic AI and asking the big question: What does Big Tech’s goal of creating a universal digital intermediary mean for society?
OMI Europe Director Max von Thun moderated a forward-looking discussion featuring Ori Schwartz from the OECD, Luca Aguzzoni from the European Commission, Frederike Kaltheuner, and Udbhav Tiwari from Signal.
The through-line across all four speakers: This technology is arriving into a tech stack already dominated by a handful of gatekeepers – and existing competition frameworks, designed to look at markets one at a time, aren't built to catch that. Big Tech wants agentic AI to become people’s gateway to the internet, concentrating market power, reshaping the digital economy, and potentially making the open web (even more) structurally obsolete. Avoiding this outcome, which will harm consumers, small businesses, and democracy, will require an ambitious, whole-system regulatory approach.
Ori Schwarz explained how, by creating a new digital intermediation platform, agentic AI risks generating significant and deeply problematic gatekeeper dynamics. He outlined why the traditional competition policy approach of looking at markets in isolation is inadequate and outdated when it comes tech markets, and particularly AI.
Luca Aguzzoni, representing the Commission, noted that they are well aware that existing gatekeepers will not shy away from leveraging dominant positions – and that this question should be central to the agenda of any enforcement authority.
Frederike Kaltheuner warned of the broad societal implications of the agentic AI market being controlled by small number of players. She set out why Europe’s AI industrial policy needs to take a full view of concentration in the AI stack and the dependencies it creates.
Udbhav Tiwari, from a third-party developer’s perspective at Signal, explained how agentic AI, imposed by dominant gatekeepers with no opt-out options for developers of third-party apps, threatens to undermine their business models and the privacy protections they offer. Operating systems must allow developers and users to make some apps off-limits to agentic AI.
Our takeaway: agentic AI’s power concentrating threat cannot be addressed in isolation. It is emerging within an already concentrated tech ecosystem that gives dominant firms significant structural advantages. A bold and ecosystem-level approach to market design is the only way regulators will be able to ensure agentic AI assistants are a technology that works for people and small businesses, not just Big Tech.