Google is burying the web under its own AI answers
Google's auto-expanding AI Overviews push organic results further down—shifting the search-ad ecosystem in ways that could permanently alter how the web surfaces knowledge.
Google is eating its own lunch, and the internet is along for the ride.
Google is quietly making a change that matters far more than most people realize. For some searches, its AI Overview at the top of the results page now auto-expands—full, dense, self-contained—before the first organic link ever appears below it. The effect is not subtle. It pushes the familiar blue-link list much farther down the page, into territory most users never reach without deliberate scrolling.
A Google spokesperson, Jennifer Kutz, told The Verge this applies “for some queries” where systems determine it is “most useful for people.” Research cited by the company claims users find Search more helpful and engage deeper in follow-up exploration. The caveat: if you have already started scrolling past the AI block, expansion cancels so you do not lose your reading position. It is a polite acknowledgment that the default experience is aggressive.
The auto-expand behavior is inconsistent even within Google’s own testing. A query like “what is the best game on Switch 2” reliably triggers it; other searches do not. Criteria are unclear. But the direction is not ambiguous. The AI block is growing, and the organic block is shrinking in view.
What this does to the ad ecosystem
The structural shift here is from distribution to destination. Search advertising has always relied on the idea that users see multiple results before clicking through. Even the top ad competes with four or five organic links for attention. Auto-expanding AI Overviews change that math dramatically. A user who gets a complete answer from the AI block may never scroll past it. They have reached their intent. The clicks disappear.
This is not theoretical. Google’s own AI Overview feature has been compressing organic traffic for months, even before auto-expand. Publishers have documented revenue declines. The effect concentrates on informational queries—how-to content, comparisons, product research—the very queries that historically generated the highest intent clicks and the strongest CPMs. Transactional queries remain closer to the fold; users still click for price checks and purchase paths. But the discovery layer, the fuzzy top of funnel where brands surface for the first time, is being hollowed out.
Who wins, who loses
Google wins. Every auto-expanded AI Overview is a moment where the user stays inside Google instead of leaving the property. Engagement metrics improve. Ad impressions in the traditional sense may decline, but Google controls the economics here. It can introduce sponsored placements inside AI Overviews themselves, or restructure attribution to reward its own inventory. The platform captures value that previously leaked to third-party publishers.
Content creators and publishers lose. Not uniformly—large brands with diversified traffic channels will absorb the hit. Independent publishers, niche review sites, mid-market content operations, those whose business models depend on organic search volume for discovery, feel the pressure immediately. A single auto-expand event can drop a piece from ten thousand daily visitors to two thousand overnight. The math is brutal.
Users appear to win. They get answers without clicking through three pages. Convenience improves. But convenience has a cost: the web becomes less discoverable. When AI Overviews become the default entry point to information, the surface area of the web contracts. Links that used to carry new audiences shrink. New publishers lose their onboarding ramp. The long tail of knowledge, the obscure corners where unexpected connections happen, gets buried under increasingly polished synthetic summaries.
The hidden pricing-out
The most consequential effect of this change is not the traffic loss itself. It is what traffic loss enables: a structural pricing-out of organic discovery. When every informational query can be answered by Google without a click, the economic justification for investing in external content weakens. Brands stop funding deep research. Publishers stop hiring writers. The supply of fresh, independent knowledge contracts precisely at the moment Google needs it most—because Google trains its models on the same web it is now suffocating.
This creates a feedback loop. Less external content means lower-quality training data. Lower-quality training data means AI Overviews become more generic and less useful. More generic results push users back toward scrolling for real answers. But by then, the publishers who could have supplied those answers have already pivoted or exited. The ecosystem has narrowed.
What happens next
The inconsistency of auto-expand matters less than its trajectory. Google introduced AI Overviews gradually, testing criteria, monitoring backlash, adjusting pacing. Auto-expand follows the same pattern. Once the feature proves engagement gains without causing measurable user friction, the criteria will broaden. Queries that currently show the compact “Show more” preview will begin expanding automatically. The wall of text at the top of results will become the standard, not the exception.
Google will likely respond to publisher pressure by introducing paid placements inside AI Overviews—a sponsored footnote, a branded suggestion embedded in the synthetic summary. The company has signaled this direction before. When it arrives, the distinction between organic and paid will collapse entirely. Every answer the user sees will carry some commercial signal. The concept of an unbiased discovery path becomes fiction.
For the web, the implication is either adaptation or decline. Publishers that diversify beyond search—newsletters, social distribution, direct traffic, subscription models—will survive. Those that do not will face a gradual revenue squeeze that makes investment in content untenable. The winners will be platforms that own the user relationship directly. The losers will be intermediaries whose value depended on being the bridge between question and answer. Google is removing the bridge and selling the destination.
The auto-expand feature is small in isolation. One additional click, one longer summary, one more scroll past the blue links. But it is a signal of the architecture Google is building. Search is becoming a product, not a portal. The internet is no longer the default place to find answers. It is an optional resource, consulted only when Google’s own summary proves insufficient. That shift changes everything about how knowledge surfaces, who profits from it, and what the web looks like five years from now.