Google, Amazon, Microsoft and Meta could extend their dominance into AI agents, the Open Markets Institute warns in a 71-page report published in November 2026. The report, 'Same Gatekeepers, New Gateway: AI Agents, Monopoly Power, and Democratic Risk', argues the four firms can carry control of cloud, models, operating systems, browsers, search and social platforms into agents: assistants that choose products, complete purchases, book travel and write messages. Its proposals are advocacy, not rules, and the report presents no new data of its own.
The report's argument rests on a simple observation: AI agents do not run in a vacuum. They run on cloud infrastructure, on foundation models, inside operating systems and browsers, and on top of search and social platforms. Google, Amazon, Microsoft and Meta control, or hold strong positions in, most of those layers between them. The report argues that control of the layers underneath gives these firms the power to shape what agents can do, which agents users encounter first, and whose products and services an agent is most likely to choose.
Agents change the stakes because they act on the user's behalf. A search engine shows ten links and the user picks one. An agent chooses a product, completes a purchase, books travel or writes a message, often without the user ever seeing the alternatives. Whoever controls the agent, the report argues, controls the choice, and whoever controls the layers the agent runs on can tilt that choice toward their own services. The gatekeeper does not disappear. The gate moves from the search results page into the agent itself.
The report sets out a series of proposals, and it is important to be clear about their status: these are advocacy positions from a think tank, not rules, laws or enforcement actions. The proposals include loyalty duties requiring agents to act in their users' interests, guaranteed access for rival agents to platforms and services, easy switching between agents, and, in some cases, forced separation of cloud businesses from AI businesses. Each is presented as a way to keep the agent layer open rather than letting it be captured by the firms that own the layers below it.
Whether any of these proposals becomes policy is an open question, and the report does not claim otherwise. Its value for readers is as a framing document: it names the competition questions that regulators, publishers and the SEO industry will be arguing about as agents become a real distribution channel. The proposals are the opening bid in that argument, not its conclusion, and they should be read as such.
The report places its argument in a regulatory context that is already moving. It notes the EU probe opened on December 9, 2025, into Google's use of publisher content for AI Overviews and AI Mode, a case that goes directly to whether AI features can draw on publishers' work without sending traffic back. It also cites the UK Competition and Markets Authority's Publisher Conduct Requirement of June 3, 2026, which addresses how Google treats publishers in its AI products. Both are live regulatory fronts, not hypotheticals.
That backdrop matters because it shows the agent question is arriving on top of an unresolved AI search question. Regulators are already examining how AI Overviews and AI Mode use publisher content and whether publishers get a fair deal. Agents add a further layer: if an agent reads publisher content, chooses a product and completes a purchase without the user visiting a site, the publisher and the merchant are both further from the transaction. The report's argument is that the same concentration of power explains both problems, and that the remedies should be designed together.
If agents become the way people choose products and services, visibility shifts from ranking to recommendation. A page that ranks first in search but is never chosen by an agent may as well not exist, while a brand the agent trusts gets picked without a search ever happening. For SEOs, that means the work of being findable, accurate structured data, clear product information, consistent brand signals across the web, becomes work of being choosable. The metrics will change, but the underlying discipline of making your content the easiest trustworthy answer does not.
For link builders, the report is a reminder to watch the agent layer as a distribution channel in its own right. Links and mentions are part of how agents learn which brands and products to trust, and the sources agents read will shape their choices. Building a brand presence that agents encounter, through citations, reviews, authoritative content and genuine mentions, is the link-building work of the agent era. The policy fight will take years. The visibility shift is already under way.
Sources: PPC Land, November 9, 2026, reporting on the Open Markets Institute report 'Same Gatekeepers, New Gateway: AI Agents, Monopoly Power, and Democratic Risk' (November 2026).
Related reading on this site: Türkiye Opens Full Antitrust Probe Into Google AI Overviews and AI Mode, Google Builds CMA-Mandated AI Opt-Out as Publishers Await Click Data, Brave Sells Its Search Index to AI — and the New York Times Is Suing, Google's Use of Press by AI Chatbot May Infringe Copyright Rules, EU Court Adviser Says, Cloudflare CEO Likens AI Agent Optimization to a New Version of SEO, and Google Is Emailing Sites: Site Reputation Penalties No Longer Apply in the EEA.
Linkbaits helps you create content that earns citations and backlinks. Start with the toolkit and build your first linkable asset this week.
Start free trial