When Google launched its AI Overviews feature — the generative summaries that now sit atop search results for hundreds of millions of queries — the company pitched it as a way to synthesize the best information the web had to offer. A smarter, faster way to get answers. What Google didn’t emphasize was just how often those AI-generated answers would point users right back to Google’s own products.
That’s the central finding of a recent investigation by WIRED, which analyzed over 300 AI Overview responses across a range of search categories — health, travel, shopping, local business, and general knowledge queries. The results were striking: roughly 42% of AI Overviews contained at least one link or reference to a Google-owned property. Google Maps. YouTube. Google Flights. Google Shopping. The Play Store. The company’s AI wasn’t just summarizing the web. It was funneling users deeper into Google’s own infrastructure.
Not subtly, either. In some categories, the self-referencing was near-total. Travel queries, for instance, returned AI Overviews that linked to Google Flights or Google Hotels in the vast majority of cases. Shopping queries frequently surfaced Google Shopping results or product listings hosted within Google’s own merchant tools. For local business searches, Google Maps was omnipresent.
This matters enormously.
It matters because AI Overviews have rapidly become the most prominent feature on Google’s search results page, often consuming the entire visible screen on mobile devices before a user scrolls. According to data from search analytics firms, AI Overviews now appear on a significant and growing share of all Google searches. When the AI summary is the first — and sometimes only — thing a user reads, the sources it cites carry outsized influence. And when those sources disproportionately belong to Google, the competitive implications are severe.
Publishers and independent websites have been sounding alarms about this dynamic for months. The concern isn’t hypothetical. Multiple analyses from SEO research firms, including studies by BrightEdge and Authoritas, have shown that AI Overviews are already reducing click-through rates to third-party websites. Users get their answer from the AI summary and never scroll down. When they do click a link, it’s increasingly one that leads to a Google product rather than an independent site.
Google has pushed back on this characterization. A company spokesperson told WIRED that AI Overviews are designed to help users “find what they’re looking for” and that the feature drives “meaningful traffic” to a diverse range of websites. The company has also pointed to internal data suggesting that links within AI Overviews receive higher click-through rates than traditional search results. But those claims are difficult to verify independently, and they sidestep the structural question: whether Google is using its dominance in search to preference its own products within AI-generated answers.
That structural question has a name in antitrust law. Self-preferencing.
It’s the same behavior that earned Google a €2.4 billion fine from the European Commission in 2017 over its Google Shopping service. In that case, regulators found that Google systematically displayed its own comparison shopping service more prominently than competitors’ offerings in search results. The company has appealed elements of that ruling, but the European Court of Justice upheld the core finding in September 2024. Google used its search dominance to give its own products an unfair advantage. Full stop.
Now the same pattern appears to be repeating — but with a new and more powerful mechanism. AI Overviews don’t just rank Google’s products higher in a list of blue links. They weave Google’s products directly into AI-generated prose that carries an implicit authority. When an AI summary says “flights from New York to London start at $350” and the only clickable link goes to Google Flights, users aren’t choosing Google Flights over Kayak or Expedia. They’re being guided there by an AI system Google itself controls.
The antitrust pressure is building from multiple directions simultaneously.
In August 2024, U.S. District Judge Amit Mehta ruled that Google holds an illegal monopoly in the search market, a landmark decision in the Justice Department’s antitrust case against the company. The remedies phase of that case is now underway, and the DOJ has floated proposals that could include forcing Google to share search data with competitors or even divesting the Chrome browser. The introduction of AI Overviews — and their apparent tendency to favor Google’s own properties — adds a new dimension to the government’s argument that Google uses its monopoly position to entrench its dominance across adjacent markets.
The European Union is watching closely too. The Digital Markets Act, which took full effect in March 2024, explicitly prohibits designated “gatekeepers” — a category that includes Google — from self-preferencing their own services in search results. Investigations under the DMA could move faster than traditional antitrust proceedings, with the European Commission empowered to impose fines of up to 10% of global annual revenue for violations. Margrethe Vestager’s successor as competition chief, Teresa Ribera, has signaled that enforcement of the DMA’s self-preferencing rules will be a priority.
And it’s not just regulators. Competitors are taking notice. Yelp, which has long accused Google of burying its restaurant and business reviews in favor of Google’s own local listings, filed an antitrust lawsuit against Google in 2024. The company’s CEO, Jeremy Stoppelman, has argued that AI Overviews represent an escalation of the same anti-competitive behavior Yelp has been fighting for over a decade. When an AI summary answers a query like “best Italian restaurants near me” by pulling in Google Maps listings and Google reviews — with no mention of Yelp, TripAdvisor, or other review platforms — the effect on competitors is tangible.
The publishing industry faces perhaps the most existential version of this threat. News organizations, recipe sites, health information providers, and countless other content creators have spent years producing the material that Google’s AI now synthesizes. Their reward? Declining traffic as users consume AI-generated summaries without clicking through. Some publishers have begun blocking Google’s AI training crawlers, but the calculus is brutal: block Google and lose search visibility entirely, or allow Google to train on your content and watch your traffic erode anyway.
This tension exploded into public view in early 2025 when several major media companies, including News Corp and the New York Times (already in litigation with OpenAI), began publicly questioning whether Google’s AI Overviews constitute a form of content appropriation. The argument is straightforward. Google’s AI reads their articles, synthesizes the information, presents it to users in Google’s own interface, and often links back to Google’s own products rather than the original source.
Google’s defense rests on a few pillars. First, the company argues that AI Overviews are fundamentally a search feature, not a content product — they help users find information, not replace the sources of that information. Second, Google points to the links embedded within AI Overviews as evidence that the feature drives traffic to publishers. Third, the company notes that users can collapse or skip AI Overviews to access traditional search results.
Each of these arguments has weaknesses. The distinction between a “search feature” and a “content product” blurs considerably when the AI generates multi-paragraph summaries that directly answer a user’s question. The traffic argument is undermined by independent data showing declining click-through rates. And the ability to skip AI Overviews is largely theoretical — the feature is enabled by default, prominently displayed, and designed to capture attention before anything else on the page.
There’s a deeper irony here. Google built its empire by organizing the world’s information and making it accessible. The implicit bargain was that Google would send traffic to the sites that created that information. AI Overviews threaten to break that bargain — not by eliminating links to external sites, but by making those links less necessary. And when the AI does link somewhere, it increasingly links home.
So where does this go? The regulatory machinery is in motion but slow. The DOJ’s remedies in the search monopoly case likely won’t take effect for years, assuming they survive appeals. The EU’s DMA enforcement is faster but still in early stages. Congressional action in the U.S. remains unlikely in the near term given the current political environment’s focus on other tech issues, particularly around TikTok and content moderation.
Market dynamics may force the issue before regulators do. If AI Overviews continue to erode traffic to independent sites, the quality of the web’s information base will degrade. Google’s AI can only synthesize what exists. If publishers stop producing high-quality content because the economics no longer work, the AI’s outputs will suffer. Google knows this, which is why the company has been quietly negotiating licensing deals with select publishers — a tacit acknowledgment that the current arrangement isn’t sustainable.
But licensing deals with a handful of major publishers won’t solve the structural problem. The long tail of the web — the small sites, niche blogs, independent experts, and community forums that collectively represent the internet’s greatest informational asset — can’t negotiate individual deals with Google. They’re simply being absorbed.
What WIRED’s analysis reveals isn’t a bug. It’s the logical outcome of a system in which the same company controls the search engine, the AI that summarizes search results, and many of the products those summaries recommend. Google has built a machine that feeds itself. The question now is whether anyone — regulators, competitors, or the market itself — can change the wiring before it becomes permanent.
WebProNews is an iEntry Publication