When the big tech companies build advanced AI directly into their search engines, it creates a huge mess for regulators and smaller competitors. This is changing how search results are made and shown, raising serious questions about fair competition and market dominance that our old AI antitrust rules just weren’t built for. The real problem is figuring out how to stop these giants from killing off innovation and consumer choice before it’s too late.
Key Takeaways
- Regulators are watching AI search like hawks for self-preferencing and data hoarding, and the US Department of Justice has made it clear they’re stepping up enforcement.
- Tech giants get a massive leg up by plugging their own AI models into search, building a walled garden that locks out third-party developers.
- A good antitrust strategy has to be proactive, pushing for things like interoperability and data access *before* the market is permanently locked down.
- Past attempts to regulate search got it wrong. They were busy policing traditional web indexes and were completely blindsided by how generative AI would change the game.
- Success would look like a 15% jump in new search engine startups within five years and a 10% drop in how much users rely on a single dominant platform.
“Meta recently updated its commercial terms of use, declaring that the company can “use any content from [a business account’s] domains, web pages, or URLs to provide, develop, improve, research, train, and maintain the integrity of” Meta products and services.”
The Looming Shadow of Algorithmic Dominance
We’ve been dealing with the power of a few tech companies in search for years, but the fast rollout of sophisticated AI is pouring gasoline on that fire in ways our old antitrust tools can’t handle. The issue goes deeper than just market share. It’s about the very architecture of how we find information. When a dominant search engine uses its own proprietary large language model (LLM) to answer your question directly and bypasses the list of links, it completely rewrites the rules of the game. The goalpost has moved from ranking web pages to being the definitive answer machine, a position that gives it unheard-of control over the flow of information.
Think about what this does to a small business or content creator. If a user’s question gets a complete answer from an AI summary right on the search page, why would they ever click through to an external site? This “zero-click” problem was already a trend, but now it’s accelerating. A local business that depends on search traffic could suddenly become invisible if an AI assistant just scrapes and summarizes its information instead of sending users to its website. It’s a vicious cycle: the dominant platform gets more data from people using its AI, which it uses to make its models even better, while competitors and publishers get starved of the traffic they need to even survive. The EU is already trying to tackle some of these gatekeeper issues with its Digital Markets Act (DMA), which came into full effect in March 2024, but AI introduces a whole new level of complexity.
What Went Wrong: Traditional Antitrust’s Blind Spots
Previous attempts to rein in big tech in search mostly focused on things like self-preferencing in rankings or exclusive ad deals. These were valid concerns, but they completely missed the structural earthquake AI was about to cause. Take the European Commission’s 2017 antitrust case against Google over its shopping service. That entire fight was about how Google displayed competing services in its search results, a battle fought over the old-world territory of web links and ranking algorithms.
The big mistake in these earlier interventions was the failure to see that search was about to pivot from a directory of “links” to an engine of “answers.” Regulators were always a step behind, reacting to abuses that had already happened instead of proactively setting the rules for new technologies. They were focused on fair access to the existing search index, but AI-powered search is moving past that index to create entirely new content. The very definition of a “search result” is changing, and so are the opportunities for anti-competitive behavior. We got a preview of this with early voice assistants, where default settings always pointed you to the manufacturer’s own services. The US Federal Trade Commission (FTC) has also admitted how hard it is to apply existing laws to tech markets that evolve faster than litigation can move.
Another blind spot was underestimating just how powerful data is as a competitive moat. Everyone knew data was an advantage, but nobody in policy circles really grasped the insane scale and quality of data needed to train a top-tier AI model. This has left us in a situation where only a couple of companies have the massive, diverse datasets needed to build and improve these LLMs, making it nearly impossible for anyone new to compete. This isn’t just about user data, either. It’s about the proprietary training data, often scraped from the entire open web, that is the foundation for these systems.
A Multi-pronged Solution for AI-Driven Search Competition
Fixing the unique problems of AI antitrust in search demands a forward-looking strategy that goes way beyond the old playbook. We need a mix of regulatory foresight, mandated interoperability, and solid data access policies.
1. Proactive Regulatory Frameworks and Market Monitoring
Instead of waiting for a monopoly to form, regulators need to set clear rules for AI deployment in search right now. That means defining what fair play looks like when a company embeds its own AI into its core search product. The US Department of Justice (DOJ) is already signaling its interest here, with Assistant Attorney General Jonathan Kanter arguing that current antitrust laws are flexible enough to work, but enforcement has to be sharp and fast. This proactive approach should include regular market studies looking specifically at AI’s effect on competition to spot choke points before they become permanent.
A huge area of concern is the “stacking” of AI capabilities. When one company owns the cloud infrastructure, the foundational AI models, and the search engine itself, it creates a vertical monopoly that can easily squeeze out everyone else. Regulators have to examine whether this vertical integration is being used for unfair self-preferencing or to erect barriers that small AI developers can’t overcome. For instance, if a search engine’s AI consistently cites its own services over better third-party options, that’s a five-alarm fire. The UK’s Competition and Markets Authority (CMA) is already digging into these dynamics with its reports on AI foundation models.
2. Mandating Interoperability and API Access
A critical move is to mandate open API access and interoperability standards for foundational AI models and search functions. This would let third-party developers build their own competing AI search tools or plug their specialized AI models into existing platforms. Can you imagine a world where a niche AI model that’s brilliant at answering medical questions could plug its results directly into a major search engine, giving users a real alternative to the platform’s generalist AI? This would create a much more dynamic field of specialized AI agents and services.
The EU’s Digital Markets Act (DMA) gives us a starting point by requiring “gatekeepers” to let third-party services interoperate with theirs. That same principle has to be explicitly applied to the AI models and data feeds that power modern search. Specific rules could require dominant search companies to offer standardized APIs for their AI summaries, letting other services display them or even swap in their own summary engines. This isn’t about breaking up companies. It’s about prying open the infrastructure that everyone needs for AI innovation.
3. Data Portability and Access for Training AI
Data is the food that AI eats. To level the playing field, we need policies that promote data portability and provide controlled access to anonymized, aggregated datasets. This isn’t about forcing companies to give away their secret sauce, but about giving competitors a fair shot at training their own models. One idea is to mandate data trusts or secure data enclaves where (with strict privacy protections) anonymized user interaction data from dominant platforms could be used to train competing AIs. This would directly attack the data moat problem.
On top of that, policy should prevent the exclusive licensing of huge datasets that are essential for training AI. If one tech giant buys up the exclusive rights to, say, a massive archive of scientific papers or medical records, it starves competitors of the raw material they need to build. Regulators should be scrutinizing these deals and could even require non-exclusive licensing for publicly funded data that’s critical for AI development. The work being done by the US National Institute of Standards and Technology (NIST) on its AI risk management framework could provide a good foundation for these kinds of data governance rules.
Measurable Results: A More Dynamic Search Ecosystem
Putting these solutions in place would bring real, measurable changes to market competition. We should see a tangible increase in the number of viable AI-powered search alternatives hitting the market. A realistic goal would be a 15% increase in new search engine market entrants over the next five years, which would be a clear sign of a healthier competitive environment. It’s not about dethroning the incumbents overnight, but about giving people real choices.
Users would get more innovation and diversity in their search results. Instead of a one-size-fits-all AI answer, you could pick from specialized AI agents built for specific tasks, leading to better and more reliable information. We could measure this with a 10% reduction in average user reliance on a single dominant search platform, tracked through things like browser defaults and usage stats. That would show a real shift away from today’s concentrated market toward a multi-platform habit.
For developers and publishers, the change would be just as big. With mandated interoperability and data access, small AI startups would have a fighting chance to get to market, which would fuel a wave of innovation that’s currently getting smothered. You’d see it in the money: a 20% increase in venture capital funding for AI search startups that aren’t tied to the big tech giants would prove that investors believe in a more open market. This would also open up new ways to make money beyond just ads, as specialized AI services could offer paid premium features and create new business models that are impossible right now.
This pressure would also force the existing giants to innovate more seriously, not just to hold their position but to actually stand out in a crowded field. That means better AI models, more transparency about data, and paying more attention to what users want instead of just coasting on network effects and default settings. The point isn’t to punish success. It’s to make sure success is won through genuine innovation in an open market, not by using a monopoly.
Conclusion
The rise of AI in search is a direct challenge to our competition policies, and it demands a smart, proactive response. By focusing on interoperability, data access, and close market monitoring, regulators can make sure the power of AI helps everyone, not just the handful of companies that got there first. The time to get this right is now, before the algorithmic gates slam shut for good.
What is AI antitrust in the context of search?
It’s about applying competition law to dominant tech companies that integrate AI into their search products. The main concerns are self-preferencing their own services, monopolizing the data needed to train AI, and creating barriers that make it impossible for new companies to compete.
How do AI-powered search engines create antitrust concerns?
They raise red flags by using AI-generated content to answer queries directly on the results page, which reduces traffic to external websites and hurts smaller publishers. They also rely on massive proprietary datasets and computing power, creating a huge barrier to entry for any competitor without those same resources.
What are “zero-click” results and why are they relevant to AI antitrust?
A “zero-click” result happens when your question is answered right on the search results page, so you don’t need to click to another website. AI is making this happen more and more. It’s an antitrust issue because it can starve third-party sites of traffic, make them less visible, and cement the dominant search platform’s role as the one-stop-shop for information.
What specific regulatory actions are being considered for AI in search?
The main ideas include forcing companies to create interoperability standards for their AI models, requiring open API access to their core AI services, and enacting data portability rules. Regulators are also looking hard at vertical integration, where one company controls every part of the AI and search stack. Authorities like the EU (with its Digital Markets Act) and the US DOJ are already looking into these options.
How can consumers benefit from increased competition in AI search?
More competition would lead to more diverse and creative search tools, like specialized AI agents for specific topics (e.g., medical or legal), better privacy protections, and more transparency in how AI generates answers. Real competition forces everyone to constantly improve their product instead of letting one company control how we access information.
